Linux Critique CVSS 9.8
46 failles dans le noyau Linux en une semaine — votre branche a déjà le correctif, mais pas votre noyau
Le projet Linux a publié 46 CVE entre le 2 et le 8 août 2026. Toutes sont déjà corrigées dans les branches stables, aucune n’est un 0-day, mais l’écart entre le correctif disponible et le noyau qui tourne sur vos machines reste le vrai risque.
2 août 2026. 8 août 2026. 46 CVE. En sept jours, le projet Linux a publié 46 avis de sécurité couvrant l’ensemble des branches stables, de 5.10 LTS à 7.2-rc5 mainline. Aucun 0-day. Aucun exploit public. Toutes les failles sont déjà corrigées — et pourtant, la plupart des noyaux en production ne les ont pas encore reçues.
La semaine du 2 au 8 août 2026 est un bon résumé de ce qu’est devenu le rythme de la sécurité kernel : un volume élevé de correctifs à faible bruit, où la seule décision opérationnelle qui compte est la vitesse à laquelle vous déployez la mise à jour de votre branche.
Une action unique : monter de version
La stratégie n’a pas changé. Exécutez uname -r, comparez avec la cible de votre branche, mettez à jour si vous êtes en dessous. Chaque correctif de la semaine est inclus dans le point release indiqué.
| Branche stable | Mettre à jour vers | Note |
|---|---|---|
| 5.10 LTS | 5.10.264 | Affectée ; le correctif SRSO CVE-2026-68480 fixe ce point release. |
| 5.15 LTS | 5.15.215 | Affectée, identique à 5.10. |
| 6.1 LTS | 6.1.182 | Correctifs backportés de la semaine. |
| 6.6 LTS | 6.6.150 | Release courante. |
| 6.12 LTS | 6.12.102 | Release courante. |
| 6.18 LTS | 6.18.43 | Release courante. |
| 7.1 stable | 7.1.7 | Release stable actuelle. |
| mainline | 7.2-rc5 | Tous les correctifs mergés. |
Les séries 6.19 et 7.0 sont en fin de vie — migrez ces systèmes vers 7.1.7 sans attendre.
Les bugs atteignables par le réseau : priorité absolue
Toutes les CVE ne se valent pas. Cinq failles sont atteignables over-the-air ou via le réseau sans authentification préalable — ce sont elles qu’il faut planifier en premier.
CVE-2026-64571 — lecture hors limites dans le pilote Wi‑Fi p54. Une trame craftée pendant la lecture EEPROM lit au-delà du socket buffer. Atteignable sans association au réseau sur tout périphérique utilisant un adaptateur p54 USB ou PCI (CONFIG_P54_COMMON).
CVE-2026-64573 — écriture hors limites dans le parseur NVM Bluetooth Qualcomm. Un underflow de longueur permet d’écrire une structure de 12 octets au-delà d’un buffer firmware court. Les puces Qualcomm BT équipent la majorité des smartphones et casques automobiles (CONFIG_BT, CONFIG_BT_QCA).
CVE-2026-64564 — use-after-free dans SCTP (CONFIG_IP_SCTP). Atteignable depuis le réseau, SCTP transporte la signalisation dans certaines stacks télécom et cliniques (DICOM/HL7).
CVE-2026-64577 — buffer under-run et panic dans le chemin GTP-U echo (CONFIG_GTP). Atteignable depuis le réseau, concerne les passerelles de cœur de réseau mobile.
CVE-2026-64578 — lecture hors limites dans le serveur SMB kernel (ksmbd). Un client SMB peut déclencher la faille via une requête compound forgée (CONFIG_SMB_SERVER). Tout NAS ou passerelle exposant un partage SMB kernel est concerné.
Virtualisation : trois bugs guest-to-host
Les hébergeurs et clouds privés doivent planifier les trois failles KVM de la semaine. Elles concernent les hôtes exécutant de la virtualisation imbriquée sur processeurs Intel (CONFIG_KVM_INTEL avec mode nested activé).
CVE-2026-64562 — use-after-free où un shadow VMCS est libéré alors qu’il est encore référencé et qu’une migration de vCPU est en cours avec un VMCLEAR.
CVE-2026-64561 — l’activité d’un invité mappe des pages dans une racine shadow-MMU invalide.
CVE-2026-68081 — fuite de pages hôtes pinnées lorsqu’un VM-Enter imbriqué échoue.
Ces trois bugs ne nécessitent pas de privilèges dans l’invité. Un locataire malveillant sur un hôte mutualisé peut potentiellement lire la mémoire d’un autre invité ou provoquer un déni de service.
Kernel config : votre meilleur filtre
Chaque CVE de cette semaine est conditionnée par une option de configuration kernel. Si CONFIG_MAC80211, CONFIG_P54_COMMON, CONFIG_BT, CONFIG_SMB_SERVER ou CONFIG_IP_SCTP ne sont pas compilés, les bugs correspondants ne peuvent pas vous affecter.
# Vérifiez en une commande les fonctionnalités exposées
zcat /proc/config.gz | grep -E 'CONFIG_(BT|MAC80211|P54_COMMON|SMB_SERVER|IP_SCTP|KVM_INTEL|BRCMFMAC)'Cette commande vous donne en cinq secondes la liste des CVE qui vous concernent réellement. Sur un serveur headless minimal, aucune des failles Wi‑Fi ou Bluetooth ne s’applique — vous passez de 46 CVE à une dizaine.
Médical, embarqué, automobile : chaque secteur a son sous-ensemble
Mobile et automobile. CVE-2026-64586 (use-after-free Broadcom FullMAC Wi‑Fi) et CVE-2026-64573 (Bluetooth QCA) sont les deux failles les plus probables sur ces plateformes. Les bugs mac80211 CVE-2026-64568 et CVE-2026-64570 concernent les points d’accès logiciels et les configurations multi-link.
Embarqué et IoT. CVE-2026-64567 (lecture hors limites dans le free-space cache btrfs) affecte les périphériques qui montent des médias amovibles en btrfs. CVE-2026-64565 est un buffer overflow heap dans le pilote IMS PCU, déclenché par un périphérique USB malveillant.
Médical. Les dispositifs connectés utilisant Bluetooth doivent intégrer le correctif CVE-2026-64573. Les équipements de lit avec périphériques USB sont concernés par CVE-2026-64565. Le correctif SCTP CVE-2026-64564 s’applique aux dispositifs transportant DICOM ou HL7/FHIR sur IP. Attention : le patching des dispositifs médicaux est encadré par la FDA (États-Unis), le règlement MDR (Europe) et la norme IEC 62304 — planifiez la mise à jour via le processus de contrôle des changements du fabricant, pas ad hoc.
Verdict
Si vous gérez des serveurs cloud ou datacenter, priorisez les trois bugs KVM (CVE-2026-64562, CVE-2026-64561, CVE-2026-68081) et la faille xfrm CVE-2026-64581 (double free atteignable par un utilisateur non privilégié via les namespaces réseau). Déployez la mise à jour de branche sur vos hyperviseurs avant la fin de la semaine.
Si vous gérez un parc mobile ou automobile, concentrez-vous sur les correctifs Wi‑Fi et Bluetooth — CVE-2026-64571, CVE-2026-64573, CVE-2026-64586.
Si vous gérez des serveurs exposant ksmbd ou SCTP, CVE-2026-64578 et CVE-2026-64564 sont vos priorités absolues : atteignables sans authentification.
La bonne nouvelle : il n’y a aucun 0-day, aucun exploit public, et chaque correctif est déjà dans la branche stable. Le risque n’est pas dans le code — il est dans le délai entre la publication du correctif et son déploiement sur vos machines.
Références
cve
Vulnérabilités liées
CVE-2026-68082In the Linux kernel, the following vulnerability has been resolved:
libceph: fix two unsafe bare decodes in decode_lockers()
decode_lockers() in cls_lock_client.c contains two bare decode operations
that allow a malicious or compromised OSD to trigger slab-out-of-bounds
reads:
1. ceph_decode_32(p) at the num_lockers field has no preceding bounds
check. ceph_start_decoding() accepts struct_len=0 as valid -- the
internal ceph_decode_need(p, end, 0, bad) always passes -- so when an
OSD sends struct_len=0, ceph_start_decoding() returns success with
p == end. The immediately following bare ceph_decode_32(p) then reads
4 bytes past the validated buffer boundary. The garbage value is
passed directly to kzalloc_objs() as the locker count.
The sibling function decode_watchers() in osd_client.c already uses
ceph_decode_32_safe() after its own ceph_start_decoding() call.
decode_lockers() was the only site using the bare variant.
2. ceph_decode_8(p) after the decode_locker() loop has no preceding
bounds check. If an OSD crafts num_lockers such that the loop
advances p exactly to end, the subsequent bare ceph_decode_8(p) reads
one byte past the validated buffer boundary. The result is passed
directly into *type, which is used as a lock type discriminator by
callers, giving an OSD-controlled one-byte OOB read with direct
influence over the lock type field.
Fix both by replacing bare operations with their safe variants:
ceph_decode_32(p) -> ceph_decode_32_safe(p, end, *num_lockers,
err_inval)
ceph_decode_8(p) -> ceph_decode_8_safe(p, end, *type,
err_free_lockers)
The goto targets differ intentionally:
err_inval: is a new label returning -EINVAL directly. It is used for
the pre-allocation failure path where *lockers is not yet allocated
and must not be passed to ceph_free_lockers().
err_free_lockers: is the existing label. It is used for the
post-allocation failure path where *lockers is allocated and must
be freed.
ret is set to -EINVAL before ceph_decode_8_safe() so that
err_free_lockers returns the correct error code on bounds violation.
Without this, err_free_lockers would return a stale ret value (0 from
the successful decode_locker() loop), silently swallowing the error.
-EINVAL is correct for both failure paths. The data received from the
OSD is structurally malformed. -ENOMEM would misrepresent the failure
class to callers and to stable@ backporters triaging error paths.
Attacker model: a malicious or compromised OSD in a multi-tenant Ceph
deployment can trigger this against any kernel client that issues the
lock.get_info class method (e.g. during RBD exclusive lock acquisition).
[ idryomov: trim changelog, formatting ] Critique CVSS 9.8 CVE-2026-64585In the Linux kernel, the following vulnerability has been resolved:
can: esd_usb: kill anchored URBs before freeing netdevs
esd_usb_disconnect() frees each CAN netdev with free_candev() inside
its per-netdev loop and only calls unlink_all_urbs(dev) afterwards.
The per-netdev private data (struct esd_usb_net_priv) is embedded in
the net_device allocation returned by alloc_candev(), so once
free_candev() has run, dev->nets[i] points to freed memory.
unlink_all_urbs() then dereferences the freed dev->nets[i] to kill the
per-netdev TX anchor (usb_kill_anchored_urbs(&priv->tx_submitted)),
clear active_tx_jobs, and reset priv->tx_contexts[].
Reorder the teardown so the anchored URBs are killed before the netdevs
are freed, matching other CAN/USB drivers in the same directory such as
ems_usb, usb_8dev and mcba_usb, which unregister, then unlink, then
free: unregister the netdevs first (which stops their TX queues), call
unlink_all_urbs(dev) once, then free the netdevs.
This issue was found by an in-house static analysis tool. Élevée CVSS 7.8 CVE-2026-64586In the Linux kernel, the following vulnerability has been resolved:
wifi: brcmfmac: drain bus_reset work on device removal
brcmf_fw_crashed() and the debugfs "reset" entry both schedule
drvr->bus_reset, whose callback recovers drvr through container_of()
and dereferences it. The removal path frees drvr (brcmf_free ->
wiphy_free) without draining the work, so a bus_reset callback pending
or running during removal can outlive drvr.
Cancellation cannot live in brcmf_detach() or brcmf_free(): the work
callback reaches teardown through the bus .reset op (PCIe
brcmf_pcie_reset -> brcmf_detach; SDIO brcmf_sdio_bus_reset ->
brcmf_sdiod_remove -> brcmf_free), so cancelling there would wait for
the running work and deadlock.
Add a per-bus mutex (bus_reset_lock) and route all arming through
brcmf_bus_schedule_reset(), which under the lock skips when the bus is
marked removing. Each bus remove entry calls
brcmf_bus_cancel_reset_work(), which under the same lock sets removing
and cancels the work. Holding the mutex across cancel_work_sync() makes
the set-removing + drain step atomic. Every producer reaches the arming
path from process context -- the PCIe firmware-halt notification runs in
the threaded IRQ handler (brcmf_pcie_isr_thread) and the SDIO hostmail
path runs from the data workqueue -- so the mutex is taken only in
sleepable contexts. Where applicable the remove entry first stops the
firmware-crash producer: on PCIe mask the mailbox and synchronize_irq;
on SDIO unregister the bus interrupt and cancel the data worker, which
also reports firmware halts through brcmf_fw_crashed(). The mutex is
initialized at bus allocation. The SDIO suspend power-off path frees
drvr through the same brcmf_sdiod_remove() and takes the same lock;
resume re-allows the work only on a successful re-probe.
Also guard brcmf_fw_crashed() against a NULL bus_if/drvr: it can fire
before brcmf_attach() wires up drvr, and it dereferences drvr
(bphy_err/brcmf_dev_coredump) before reaching the arming gate.
The bus_reset work is shared across buses, so the drain is applied to
every remove path: PCIe (the .reset op introduced by the Fixes commit),
SDIO (arms the same work through brcmf_fw_crashed()), and USB (via the
debugfs "reset" entry). cancel_work_sync() drains a running or pending
bus_reset work item before removal frees drvr, and patch 1/2 makes the
scratch-buffer release safe when reset teardown has already released
those DMA buffers.
This patch fixes the lifetime of the bus_reset work item itself. It does
not attempt to address the separate, pre-existing lifetime of the
asynchronous firmware completion started by the PCIe reset path. That
callback needs its own lifetime/ownership protocol and is being tracked
separately.
This issue was found by an in-house static analysis tool. Élevée CVSS 8.8 CVE-2026-64597In the Linux kernel, the following vulnerability has been resolved:
smb: client: fix double-free in SMB2_close() replay
A response-bearing attempt can return a replayable error and free its
response buffer. If SMB2_close_init() fails before the next send, cleanup
retains the previous buffer type and frees that response again.
Reset response bookkeeping before each attempt to prevent the stale free. Critique CVSS 9.8 CVE-2026-64598In the Linux kernel, the following vulnerability has been resolved:
smb/client: Fix error code in smb2_aead_req_alloc()
The "*num_sgs" variable is a u32 so "ERR_PTR(*num_sgs)" doesn't work.
We would have to do something similar to the previous line where it's
cast to int and then long. However, it's simpler to store the return in
an int ret variable.
This bug would eventually result in a crash when dereference the invalid
error pointer. Élevée CVSS 8.8 CVE-2026-64567In the Linux kernel, the following vulnerability has been resolved:
btrfs: reject free space cache with more entries than pages
When loading a v1 free space cache, __load_free_space_cache() takes
num_entries and num_bitmaps straight from the on-disk
btrfs_free_space_header. That header is stored in the tree_root under a key
with type 0, which the tree-checker has no case for, so neither count is
validated before the load trusts it.
The load loops num_entries times and maps the next page whenever the current
one runs out, going through io_ctl_check_crc() -> io_ctl_map_page(), which
does io_ctl->pages[io_ctl->index++]. But pages[] is allocated in
io_ctl_init() from the cache inode's i_size, not from num_entries:
num_pages = DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE);
io_ctl->pages = kcalloc(num_pages, sizeof(struct page *), GFP_NOFS);
So if num_entries claims more records than the pages can hold, io_ctl->index
runs off the end of pages[]. The write side never hits this because
io_ctl_add_entry() and io_ctl_add_bitmap() both stop once
io_ctl->index >= io_ctl->num_pages; the read side just never had the same
check.
To trigger it, take a clean cache (num_entries = <N> here), set num_entries
in the header to 0x10000, and fix up the leaf checksum so it still passes
the tree-checker. The cache inode has i_size = 65536, so num_pages is 16 and
pages[] is a 16-pointer (kmalloc-128) array. The load now tries to read
65536 entries, io_ctl->index walks up to 16, and pages[16] is read past the
array:
BUG: KASAN: slab-out-of-bounds in io_ctl_check_crc (fs/btrfs/free-space-cache.c:420 fs/btrfs/free-space-cache.c:565)
Read of size 8 at addr ffff88800c833a80 by task kworker/u8:3/58
io_ctl_check_crc (fs/btrfs/free-space-cache.c:420 fs/btrfs/free-space-cache.c:565)
__load_free_space_cache (fs/btrfs/free-space-cache.c:655 fs/btrfs/free-space-cache.c:820)
load_free_space_cache (fs/btrfs/free-space-cache.c:1017)
caching_thread (fs/btrfs/block-group.c:880)
btrfs_work_helper (fs/btrfs/async-thread.c:312)
process_one_work
worker_thread
kthread
ret_from_fork
free-space-cache.c:420 is io_ctl_map_page(), inlined into io_ctl_check_crc()
at line 565, which is why that is the frame KASAN names. The out-of-bounds
slot is then treated as a struct page and handed to crc32c(), so the bad
read turns into a GP fault.
Add the missing check to io_ctl_check_crc(), which is where both the entry
loop and the bitmap loop end up. When num_entries is too large the load now
fails like any corrupt cache: __load_free_space_cache() drops it and rebuilds
the free space from the extent tree, so a valid cache is never rejected. Élevée CVSS 7.8 CVE-2026-64568In the Linux kernel, the following vulnerability has been resolved:
wifi: mac80211: fix unsol_bcast_probe_resp double free on alloc failure
ieee80211_set_unsol_bcast_probe_resp() calls kfree_rcu() on the old
template before allocating the replacement. If the kzalloc() then fails,
it returns -ENOMEM while link->u.ap.unsol_bcast_probe_resp still points
at the object already queued for freeing. A later update or AP teardown
re-queues that same rcu_head; the second free is caught by KASAN when the
RCU sheaf is processed in softirq:
BUG: KASAN: double-free in rcu_free_sheaf (mm/slub.c:5850)
Free of addr ffff88800d06f300 by task exploit/145
...
__rcu_free_sheaf_prepare (mm/slub.c:2634 mm/slub.c:2940)
rcu_free_sheaf (mm/slub.c:5850)
rcu_core (kernel/rcu/tree.c:2617 kernel/rcu/tree.c:2869)
handle_softirqs (kernel/softirq.c:622)
The buggy address belongs to the cache kmalloc-128 of size 128
Queue the old object for kfree_rcu() only after the new one is published,
matching ieee80211_set_probe_resp() and ieee80211_set_s1g_short_beacon(). Élevée CVSS 7.8 CVE-2026-64570In the Linux kernel, the following vulnerability has been resolved:
wifi: mac80211: fix fils_discovery double free on alloc failure
ieee80211_set_fils_discovery() calls kfree_rcu() on the old template
before allocating the replacement. If the kzalloc() then fails, it
returns -ENOMEM while link->u.ap.fils_discovery still points at the
object already queued for freeing. A later update or AP teardown
(ieee80211_stop_ap()) re-queues that same rcu_head; the second free is
caught by KASAN when the RCU sheaf is processed in softirq:
BUG: KASAN: double-free in rcu_free_sheaf (mm/slub.c:5850)
Free of addr ffff88800c065280 by task swapper/0/0
...
__rcu_free_sheaf_prepare (mm/slub.c:2634 mm/slub.c:2940)
rcu_free_sheaf (mm/slub.c:5850)
rcu_core (kernel/rcu/tree.c:2617 kernel/rcu/tree.c:2869)
handle_softirqs (kernel/softirq.c:622)
The buggy address belongs to the cache kmalloc-96 of size 96
Queue the old object for kfree_rcu() only after the new one is published,
matching ieee80211_set_probe_resp() and ieee80211_set_s1g_short_beacon(). Élevée CVSS 7.8 CVE-2026-64574In the Linux kernel, the following vulnerability has been resolved:
wifi: mac80211: tear down new links on vif update error path
When ieee80211_vif_update_links() adds new links it allocates a link
container for each and calls ieee80211_link_init() (which registers the
per-link debugfs files with file->private_data pointing into the container)
and ieee80211_link_setup(). If the subsequent drv_change_vif_links() fails,
the error path restores the old pointers and jumps to 'free', which frees
the new containers but never removes their debugfs entries or stops the
links. The debugfs files survive with file->private_data dangling at the
freed container, so a later open()+read() (e.g. link-1/txpower)
dereferences freed memory in ieee80211_if_read_link(), a use-after-free.
The removal path already dismantles links correctly via
ieee80211_tear_down_links(), which removes each link's keys and debugfs
entries and calls ieee80211_link_stop(); the add path on the error branch
does not. Commit be1ba9ed221f ("wifi: mac80211: avoid weird state in error
path") hardened this same error path for the link-removal case
(new_links == 0) but left the newly-added links' teardown unaddressed.
drv_change_vif_links() can fail at runtime on MLO drivers (internal
allocation / queue / firmware command failures).
Remove the new links' debugfs entries and stop them before freeing.
BUG: KASAN: slab-use-after-free in ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
Read of size 8 at addr ffff888011290000 by task exploit/145
Call Trace:
...
ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
short_proxy_read (fs/debugfs/file.c:373)
vfs_read (fs/read_write.c:572)
ksys_read (fs/read_write.c:716)
do_syscall_64 (arch/x86/entry/syscall_64.c:94)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
...
Oops: general protection fault, probably for non-canonical address 0xdffffc000000000a
RIP: 0010:ieee80211_if_read_link (net/mac80211/debugfs_netdev.c:127)
Kernel panic - not syncing: Fatal exception Élevée CVSS 7.8 CVE-2026-64577In the Linux kernel, the following vulnerability has been resolved:
gtp: check skb_pull_data() return in gtp1u_send_echo_resp()
gtp1u_send_echo_resp() ignores skb_pull_data()'s return value. Its
caller gtp1u_udp_encap_recv() only guarantees 16 bytes (udphdr +
gtp1_header), but the pull requests 20 (gtp1_header_long + udphdr). For
a 16-19 byte echo request the pull fails and returns NULL without
advancing skb->data; execution continues, and the following skb_push()
plus the IP header pushed by iptunnel_xmit() move skb->data below
skb->head, tripping skb_under_panic().
Fix it by dropping the packet when skb_pull_data() fails.
skbuff: skb_under_panic: ...
kernel BUG at net/core/skbuff.c:214!
Call Trace:
skb_push (net/core/skbuff.c:2648)
iptunnel_xmit (net/ipv4/ip_tunnel_core.c:82)
gtp_encap_recv (drivers/net/gtp.c:701 drivers/net/gtp.c:808 drivers/net/gtp.c:920)
udp_queue_rcv_one_skb (net/ipv4/udp.c:2388)
...
Kernel panic - not syncing: Fatal exception in interrupt Élevée CVSS 7.5 CVE-2026-64578In the Linux kernel, the following vulnerability has been resolved:
ksmbd: validate compound request size before reading StructureSize2
When ksmbd validates a compound (chained) SMB2 request,
ksmbd_smb2_check_message() reads pdu->StructureSize2 without first
checking that the compound element is large enough to contain it.
StructureSize2 is a 2-byte field at offset 64
(__SMB2_HEADER_STRUCTURE_SIZE) from the start of each element.
The compound-walking logic only guarantees that a full 64-byte SMB2
header is present for the trailing element: when NextCommand is 0, len is
reduced to the number of bytes remaining after next_smb2_rcv_hdr_off. A
remote client can craft a compound request whose last element has exactly
64 bytes, so the 2-byte StructureSize2 read at offset 64 extends one byte
past the receive buffer, producing a slab-out-of-bounds read.
BUG: KASAN: slab-out-of-bounds in ksmbd_smb2_check_message (fs/smb/server/smb2misc.c:402)
Read of size 2 at addr ffff888012ae31ac by task kworker/0:1/14
The buggy address is located 172 bytes inside of allocated 173-byte region
Workqueue: ksmbd-io handle_ksmbd_work
Call Trace:
...
kasan_report (mm/kasan/report.c:595)
ksmbd_smb2_check_message (fs/smb/server/smb2misc.c:402)
handle_ksmbd_work (fs/smb/server/server.c:119)
process_one_work (kernel/workqueue.c:3314)
worker_thread (kernel/workqueue.c:3397)
kthread (kernel/kthread.c:436)
ret_from_fork (arch/x86/kernel/process.c:158)
ret_from_fork_asm (arch/x86/entry/entry_64.S:245)
Reject any compound element that is too small to hold StructureSize2
before dereferencing it. Élevée CVSS 8.2 CVE-2026-64581In the Linux kernel, the following vulnerability has been resolved:
xfrm: fix sk_dst_cache double-free in xfrm_user_policy()
xfrm_user_policy() clears the socket dst cache with __sk_dst_reset(),
i.e. the non-atomic __sk_dst_set(sk, NULL): it reads sk_dst_cache with
rcu_dereference_protected(), stores NULL and dst_release()s the old dst.
That is only safe if no other thread modifies sk_dst_cache concurrently.
For a connected UDP socket that does not hold: the transmit fast path
(udp_sendmsg -> sk_dst_check -> sk_dst_reset) resets the cache locklessly
with an atomic xchg(). A per-socket policy change racing a send can make
both sides observe the same old dst and each dst_release() it, dropping
the socket's single reference twice and freeing the xfrm_dst bundle while
it is still referenced:
BUG: KASAN: slab-use-after-free in dst_release
Write of size 4 at addr ffff88801897b6c0 by task exploit/155
Call Trace:
...
dst_release (... ./include/linux/rcuref.h:109)
xfrm_user_policy (./include/net/sock.h:2239 ./include/net/sock.h:2256 net/xfrm/xfrm_state.c:3053)
do_ip_setsockopt (net/ipv4/ip_sockglue.c:1347)
ip_setsockopt (net/ipv4/ip_sockglue.c:1417)
do_sock_setsockopt (net/socket.c:2368)
__sys_setsockopt (net/socket.c:2393)
__x64_sys_setsockopt (net/socket.c:2396)
do_syscall_64 (arch/x86/entry/syscall_64.c:94)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:121)
Reachable by an unprivileged user via a user+network namespace.
Use the atomic sk_dst_reset() so the cache is cleared and released with a
single xchg(): whichever side wins releases the dst once, the other sees
NULL and does nothing. Behaviour is otherwise unchanged. Élevée CVSS 7.8 CVE-2026-64582In the Linux kernel, the following vulnerability has been resolved:
RDMA/rxe: Fix a use-after-free problem in rxe_mmap
rxe_mmap() removes a rxe_mmap_info struct from the pending_mmaps list
and releases pending_lock while the struct's kref is still at 1:
list_del_init(&ip->pending_mmaps);
spin_unlock_bh(&rxe->pending_lock); /* ref == 1, no lock held */
ret = remap_vmalloc_range(vma, ip->obj, 0); /* walks PTEs */
[...]
rxe_vma_open(vma); /* kref_get, ref → 2 */
remap_vmalloc_range_partial() walks PTEs without any lock.
A concurrent DESTROY_CQ ioctl on another CPU calls:
kref_put(&q->ip->ref, rxe_mmap_release) /* ref 1→0 */
vfree(ip->obj) /* clears vmalloc PTEs mid-walk */
kfree(ip) /* frees rxe_mmap_info */
This yields:
1. Kernel crash, vmalloc_to_page() returns NULL when vfree wins the
per-PTE race -> vm_insert_page(NULL) → GPF in validate_page_before_insert
2. Page UAF, vmalloc_to_page() reads a stale PTE before vfree clears
it. User VMA holds a PTE to a free'd page which might eventually get
reallocated later by vmalloc which allows the attacker to get a clean
page-level UAF.
It is worth noting that even though a page-level UAF is possible given
the strong primitive, it is statistically very difficult to achieve
given the very short time window (after the last insert_page and before
the kref_get).
The call trace are as below:
Oops: general protection fault, probably for non-canonical address 0xdffffc0000000001: 0000 [#1] SMP KASAN NOPTI
KASAN: null-ptr-deref in range [0x0000000000000008-0x000000000000000f]
CPU: 0 UID: 1000 PID: 413 Comm: poc Not tainted 7.0.0-rc5-dirty #28 PREEMPT(lazy)
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014
RIP: 0010:validate_page_before_insert+0x32/0x300
Code: e5 41 57 41 56 49 89 fe 41 55 41 54 53 48 89 f3 e8 93 b5 a3 ff 48 8d 7b 08 48 b8 00 00 00 00 00 fc ff df 48 89 fa 48 c1 ea 03 <80> 3c 02 00 0f 85 7b 02 00 00 4c 8b 63 08 31 ff 4d 89 e5 41 83 e5
RSP: 0018:ffff88811b15f2f0 EFLAGS: 00000202
RAX: dffffc0000000000 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 0000000000000001 RSI: 0000000000000000 RDI: 0000000000000008
RBP: ffff88811b15f318 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffff8881181eee00
R13: 0000000000000000 R14: ffff8881181eee00 R15: ffff8881181eee20
FS: 00007b1e000f76c0(0000) GS:ffff8884268e0000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007b1e00a24ac0 CR3: 0000000116eb3000 CR4: 00000000000006f0
Call Trace:
<TASK>
insert_page+0x8f/0x190
? __pfx_insert_page+0x10/0x10
? kasan_save_alloc_info+0x38/0x60
vm_insert_page+0x2e7/0x400
remap_vmalloc_range_partial+0x212/0x3e0
remap_vmalloc_range+0x6e/0xb0
? __kasan_check_write+0x14/0x30
rxe_mmap+0x2e9/0x5d0
ib_uverbs_mmap+0x1ad/0x2c0
__mmap_region+0x12c2/0x2ad0
? __pfx___mmap_region+0x10/0x10
? __sanitizer_cov_trace_switch+0x58/0xb0
? mas_prev_slot+0x360/0x39c0
? __sanitizer_cov_trace_switch+0x58/0xb0
? mas_next_slot+0x1e5b/0x2f40
? __sanitizer_cov_trace_cmp8+0x18/0x30
? unmapped_area_topdown+0x4dd/0x610
? kfree+0x1b1/0x440
? free_cpumask_var+0x16/0x30
? __kasan_slab_free+0x7d/0xa0
? __sanitizer_cov_trace_cmp8+0x18/0x30
mmap_region+0x2e6/0x3c0
do_mmap+0xa3e/0x12a0
? __pfx_do_mmap+0x10/0x10
? __kasan_check_write+0x14/0x30
? down_write_killable+0xba/0x160
? __pfx_down_write_killable+0x10/0x10
? __sanitizer_cov_trace_cmp4+0x16/0x30
vm_mmap_pgoff+0x2d4/0x4a0
? __pfx_vm_mmap_pgoff+0x10/0x10
? fget+0x1bf/0x270
ksys_mmap_pgoff+0x40c/0x690
? __sanitizer_cov_trace_const_cmp4+0x16/0x30
? __pfx_ksys_mmap_pgoff+0x10/0x10
? __kasan_check_write+0x14/0x30
? _raw_spin_trylock+0xbb/0x130
? __pfx__raw_spin_trylock+0x10/0x10
__x64_sys_mmap+0x135/0x1e0
x64_sys_c
---truncated--- Élevée CVSS 7.8 CVE-2026-64561In the Linux kernel, the following vulnerability has been resolved:
KVM: x86: Check for invalid/obsolete root *after* making MMU pages available
Check for a "stale" page fault, i.e. for an invalid and/or obsolete root,
after making MMU pages available for the shadow MMU. If reclaiming shadow
pages zaps an in-use root, i.e. marks it invalid, then KVM will attempt to
map memory into an invalid root. On its own, populating an invalid root is
"fine", but because child shadow pages inherit their parent's role, any
children created during the map/fetch will be created as invalid pages,
thus violating KVM's invariant that invalid pages are never on the list of
active MMU pages.
Note, the underlying flaw has existed since KVM first started tracking
invalid roots in 2008 (commit 2e53d63acba7, "KVM: MMU: ignore zapped root
pagetables"), but the true badness only came along in 2020 (Linux 5.9)
with the invariant that invalid shadow pages can't be on the list of
active pages.
Note #2, inheriting role.invalid when creating child shadow pages is also
far from ideal; that flaw will be addressed separately. Élevée CVSS 8.8 CVE-2026-64562In the Linux kernel, the following vulnerability has been resolved:
KVM: nVMX: Hide shadow VMCS right after VMCLEAR
free_nested() frees the shadow VMCS while vmcs01 still points to it. But
because it is asynchronous with respect to loaded_vmcs_clear(), the vCPU
might migrate before the pointer is cleared and __loaded_vmcs_clear()
may then execute VMCLEAR.
The VMCS needs to stay attached until its explicit VMCLEAR completes, but
then it can be hidden and the page safely freed. Élevée CVSS 8.8 CVE-2026-64564In the Linux kernel, the following vulnerability has been resolved:
sctp: don't free the ASCONF's own transport in DEL-IP processing
sctp_process_asconf() caches the transport the ASCONF chunk is processed
against in asconf->transport (== chunk->transport, set once in sctp_rcv()).
For an ASCONF located through its Address Parameter by
__sctp_rcv_asconf_lookup(), that cached transport corresponds to the
Address Parameter, which need not be the packet's source address.
sctp_process_asconf_param() rejects a DEL-IP for the packet source address
(ADDIP D8, SCTP_ERROR_DEL_SRC_IP), but nothing protects asconf->transport.
A single ASCONF can therefore carry, in order:
[Address Parameter L] [DEL-IP L] [DEL-IP 0.0.0.0]
where L differs from the source. The DEL-IP for L passes the D8 check and
calls sctp_assoc_rm_peer() on the transport that asconf->transport still
points at, freeing it (RCU-deferred). The following wildcard DEL-IP then
reuses the now-dangling asconf->transport in sctp_assoc_set_primary() and
sctp_assoc_del_nonprimary_peers(): set_primary() dereferences the freed
transport (->ipaddr, ->state) and plants the dangling pointer into
asoc->peer.primary_path / active_path, and del_nonprimary_peers(), keeping
only the pointer that is no longer on the list, removes every real
transport, leaving the association with a transport_count of 0 and
primary_path/active_path pointing at freed memory.
Reject a DEL-IP that targets the transport the ASCONF is being processed
against, mirroring the existing source-address guard, so the wildcard
branch can never reuse a freed transport. Critique CVSS 9.8