tun: Require CAP_NET_ADMIN to attach to an ownerless device

In the Linux kernel, the following vulnerability has been resolved: tun: Require CAP_NET_ADMIN to attach to an ownerless device Twenty years ago I made creating a tuntap device require CAP_NET_ADMIN in commit ca6bb5d7ab22 ("[NET]: Require CAP_NET_ADMIN to create tuntap devices.") The idea was to allow the permissions on /dev/net/tun to be relaxed to 0666 and thus allow owners of tun devices assigned with TUNSETOWNER to actually open it and get at their devices. What we missed at the time was that persistent devices with *no* owner were already essentially world-consumable, and anyone who could open the device node could also attach to those. Even then, using persistent devices with no explicitly set owner — on the (false) presumption that CAP_NET_ADMIN would be required to open them — wasn't an egregious userspace bug. There are cases where it does make sense to separate the creation/configuration of the netdevice, from the process which shovels packets into it. Since then, TUNSETGROUP was added a year later in commit 8c644623fe7e ("[NET]: Allow group ownership of TUN/TAP devices."). Its commit message noted that if both uid and gid are set, both have to match (in itself a bizarre choice) — but still not that with *neither* set, no permission at all was required. Then in 2012 commit c8d68e6be1c3 ("tuntap: multiqueue support") allowed new queues to be attached to a device from the moment of its creation, with no way for userspace to close the race between creation and a subsequent TUNSET{OWNER,GROUP}. In 2024, commit 3ca459eaba1bf ("tun: fix group permission check") tried to fix things to be saner: only require the user *or* the group to match, in line with standard POSIX permissions, and require CAP_NET_ADMIN if neither the user nor group authorised the access (although it put the CAP_NET_ADMIN check first, and would wrongly have reported the capability as having been used even when it was not). That fix was reverted in commit a70c7b3cbc06 ("tun: revert fix group permission check") due to allegedly causing a regression in an selinux test suite which *expected* the attachment to succeed when authorised by none of user, group, nor CAP_NET_ADMIN: https://lore.kernel.org/netdev/CAFqZXNtkCBT4f+PwyVRmQGoT3p1eVa01fCG_aNtpt6dakXncUg@mail.gmail.com/ We should not break UAPI. But we don't have to consider an selinux test case which deliberately exercises a security flaw to be UAPI worth preserving either. To which end, I have fixed it in selinux-testsuite commit b2f894f0016b ("tests/tun_tap: set an owner on the device for the relabel tests"). So: fix the lack of CAP_NET_ADMIN check on devices which *haven't* been assigned an owner or group, while leaving the weird tun ownership model as it was (user *and* group have to match, if both are set). If someone still wants to fix the latter (which wasn't actually the reason the previous attempt was reverted), they can do it under separate cover. Permit attaching to an existing device only if an owner and/or group was set and the caller matches all that were set, or if the caller has CAP_NET_ADMIN. The only behavioural change is that a device with no owner or group now requires CAP_NET_ADMIN to attach — precisely the loophole we want to close. And put the CAP_NET_ADMIN check in the right place this time.

tcp: Fix potential UAF in reqsk_timer_handler().

In the Linux kernel, the following vulnerability has been resolved: tcp: Fix potential UAF in reqsk_timer_handler(). When TCP socket migration fails at inet_ehash_insert() in reqsk_timer_handler(), we jump to the no_ownership: label and free the new reqsk immediately with __reqsk_free(). Thus, we must stop the new reqsk's timer before jumping to the label, but the timer might be missed since the cited commit, resulting in UAF. As we are in the original reqsk's timer context, we can safely call timer_delete_sync() for the new reqsk. Let's pass false to __inet_csk_reqsk_queue_drop() to stop the new reqsk's timer.

vsock/vmci: validate packet source for connected sockets

In the Linux kernel, the following vulnerability has been resolved: vsock/vmci: validate packet source for connected sockets vmci_transport_recv_stream_cb() looks up sockets first by the full source and destination tuple, then by destination only in the bound table. The fallback can select a non-listening socket without checking whether the packet came from its stored peer. This was reproduced with two VMCI contexts. A RST from the context not stored in a TCP_SYN_SENT socket reset that socket after it was selected by the destination-only lookup. VMCI can process notification packets in bottom-half context when the socket is not owned by user context, or defer packets to a workqueue. Use vsock_check_source() after taking the socket lock in the bottom-half path, and recheck after lock_sock() in the workqueue path. Listening sockets continue to accept packets from any source. Reply with a RST addressed from the received packet before dropping a source that fails validation. This preserves the existing reset behavior for bound non-listening and concurrently closed sockets without directing the reset to a connected socket's stored peer.

smb: client: reject SetEA requests that do not fit the request buffer

In the Linux kernel, the following vulnerability has been resolved: smb: client: reject SetEA requests that do not fit the request buffer CIFSSMBSetEA() copies the caller's extended attribute value into the SMB request buffer without checking that it fits. The requirement is stated in the source but was never implemented: /*BB add length check to see if it would fit in negotiated SMB buffer size BB */ /* if (ea_value_len > buffer_size - 512 (enough for header)) */ if (ea_value_len) memcpy(parm_data->list.name + name_len + 1, ea_value, ea_value_len); The only bound applied on the way in is in cifs_xattr_set(): #define MAX_EA_VALUE_SIZE CIFSMaxBufSize ... if (size > MAX_EA_VALUE_SIZE) CIFSMaxBufSize is the full payload capacity of the buffer, so a value of exactly that size leaves no room for the SMB header, the TRANS2 parameter block, the fealist header and the EA name that are written ahead of it in the same object. SendReceive() already enforces the correct limit on this very length: if (in_len > CIFSMaxBufSize + MAX_CIFS_HDR_SIZE) but it is called after the copy has taken place. An unprivileged setxattr(2) on an SMB1 mount with a 250-byte name and a 16384-byte value writes 16384 bytes starting 345 bytes into a 16588-byte cifs_request object, ending 141 bytes past it: BUG: KASAN: slab-out-of-bounds in CIFSSMBSetEA+0xabc/0xde0 Write of size 16384 at addr ffff888003aa0159 by task init/68 __asan_memcpy+0x3c/0x60 CIFSSMBSetEA+0xabc/0xde0 cifs_xattr_set+0xd3a/0xff0 __vfs_setxattr+0x13e/0x1a0 The buggy address is located 345 bytes inside of allocated 16588-byte region Apply SendReceive()'s limit to the assembled request before the copy rather than after it, and widen the byte counters so the sum cannot wrap before it is tested. byte_count is also tested against U16_MAX, because it is stored in the 16-bit pSMB->ByteCount. That becomes reachable when CIFSMaxBufSize is raised at module load, where it may be set as high as 1024*127: with a 5-byte EA name and a 65521-byte value, count is exactly U16_MAX while byte_count is 65556, and cpu_to_le16() would truncate it to 20 and transmit a frame whose ByteCount does not match its length. Testing byte_count covers count as well, since byte_count is the larger of the two and count's only 16-bit consumer is written after this point. check_add_overflow() is evaluated first so that total_len is assigned before it is reported.

namespace: never expire a locked mount

In the Linux kernel, the following vulnerability has been resolved: namespace: never expire a locked mount Locked mounts are special. They protect the underlying files and directories from being revealed. do_umount() refuses to unmount locked mounts but shrink_submounts() doesn't. A shrinkable mount can become locked once the owner of a user namespace puts it beneath a locked mount with MOVE_MOUNT_BENEATH. The locked property now moves to the mount at the bottom making it possible to unmount the top mount. So umount() of an unlocked ancestor now expires the bottom mount first and the covered directory is revealed: move_mount(c -> x/hidden, MOVE_MOUNT_BENEATH) = 0 the cover after the lock moved: umount2(x/hidden) = 0 the holder of the lock: umount2(x/hidden) = EINVAL the root of the copy, busy: umount2(x) = EBUSY reads x/hidden/secret: "covered-by-root" So leave a locked mount alone as it dies together with its parent. A plain umount() of an unlocked mount with a locked one below it is EBUSY from now on. It's the same for any other locked child. A lazy umount still takes the whole tree. mark_mounts_for_expiry() never sees a locked mount. lock_mnt_tree() leaves a mount on an expiry list alone and nothing else puts a locked one on such a list. Add an assert for this.

net: gso: limit recursive IP-in-IP segmentation

In the Linux kernel, the following vulnerability has been resolved: net: gso: limit recursive IP-in-IP segmentation IP-in-IP GSO can re-enter inet_gso_segment() or ipv6_gso_segment() for each nested IP header. encap_level tracks header bytes, not callback depth, so a deep chain can exhaust the kernel stack. Making inet_gso_segment() stackable introduced unbounded IPv4 nesting; IPIP GSO/TSO later made the path reachable. The IPv6 stackable path was introduced separately and uses the same guard. Count IPv4 and IPv6 GSO handler entries in skb_gso_cb, initialized once per top-level GSO operation and preserved across GRE/UDP context changes. Use the existing IP_TUNNEL_RECURSION_LIMIT for both handlers. The first five entries pass, and the sixth returns -EINVAL before dispatching another GSO callback.

KVM: SVM: Preserve TLB control (i.e. pending TLB flush) on failed VMRUN

In the Linux kernel, the following vulnerability has been resolved: KVM: SVM: Preserve TLB control (i.e. pending TLB flush) on failed VMRUN Don't reset the VMCB's TLB control back to "do nothing" on a failed VMRUN, as empirical testing shows that the CPU performs the requested TLB flush if and only if VMRUN is successful, i.e. clearing TLB control on a failed VMRUN effectively drops a TLB flush. Explicitly track the need to flush all ASIDs on a per-CPU basis, as the ASID reuse condition is tied to the pCPU, not to the vCPU. As a bonus, this also obviates the need to avoid clobbering FLUSH_ALL_ASID with TLB_CONTROL_FLUSH_ASID, e.g. in svm_flush_tlb_asid(). Deliberately don't bother saving/restoring the "old" tlb_ctl on failure, in quotes because it's not exactly the old tlb_ctl, it's the tlb_ctl from after pre_svm_run(), but before updating tlb_ctl for flush_all_asids. If VMRUN fails and TLB_CONTROL_FLUSH_ALL_ASID is forced, then the next successful run of the VMCB *may* unnecessarily flush all ASIDs, which strictly speaking could result in noisy neighbor issues. However, the fact that new_asid() is already guest-triggerable, because of KVM's flawed behavior of clearing the ASID on emulated INIT, means that a guest can already trigger a flush of all ASIDs at roughly the same rate. And once KVM stops clobbering the ASID on emulated INIT, *or* assigns a static ASID to each vCPU, this flaw goes away.

KVM: arm64: Transfer the hyp stack pages out of the host stage-2

In the Linux kernel, the following vulnerability has been resolved: KVM: arm64: Transfer the hyp stack pages out of the host stage-2 fix_host_ownership() walks only the linear-map alias of each memblock region, and the per-CPU hyp stack, mapped in the private VA range for its guard page, has none. Walk each stack's VA range with the same walker.

mips: select CONFIG_WEAK_REORDERING_BEYOND_LLSC from CONFIG_EYEQ

In the Linux kernel, the following vulnerability has been resolved: mips: select CONFIG_WEAK_REORDERING_BEYOND_LLSC from CONFIG_EYEQ On I6500 CPU cores, lld and scd give no ordering guarantees (same as all other instructions). To respect the assumption that arch_cmpxchg() is fully ordered, we must inject sync instructions above and below our lld/scd loops using the already in place WEAK_REORDERING_BEYOND_LLSC infrastructure. Otherwise, bad things can happen: [ 34.054496] CPU 3 Unable to handle kernel paging request at virtual address 0000000000000000, epc == a80000080838e01c, ra == a80000080838dfc4 [ 34.054559] Oops[#1]: [ 34.069561] CPU: 3 UID: 0 PID: 170 Comm: pipe_race Not tainted 7.2.0-rc6-01553-gb73c35220968-dirty #103 VOLUNTARY [ 34.079932] Hardware name: Mobile EyeQ5 MP5 Evaluation board [ 34.085592] $ 0 : 0000000000000000 0000000000000001 0000000000000000 0000000000000000 [ 34.093616] $ 4 : a800000808ee2618 000000000b7a879d 0000000000001000 0000000000000000 [ 34.101638] $ 8 : 0000000000e3f2c9 0000000000000000 a800000808a2a9f8 0000000000000000 [ 34.109660] $12 : a8000008139ffcd8 ffffffff84080018 a80000080837fae0 7878787878787878 [ 34.117682] $16 : a800000807e82940 0000000000001000 0000000000000000 0000000000000000 [ 34.125704] $20 : a800000802920e00 a8000008139ffdf8 a800000802649400 0000000000e3f2c9 [ 34.133726] $24 : 0000000000000006 00000001200406e0 [ 34.141783] $28 : a8000008139fc000 a8000008139ffd10 0000000000e3f2c8 a80000080838dfc4 [ 34.149837] epc : a80000080838e01c anon_pipe_read+0xd4/0x428 [ 34.155697] ra : a80000080838dfc4 anon_pipe_read+0x7c/0x428 [ 34.161549] Status: 140000e3 KX SX UX KERNEL EXL IE [ 34.166551] Cause : 40800408 (ExcCode 02) [ 34.170574] BadVA : 0000000000000000 [ 34.174161] PrId : 0001b028 (MIPS I6500) [ 34.178183] Process pipe_race (pid: 170, threadinfo=000000005ca35720, task=00000000e1013890, tls=000000014ebbb780) [ 34.188568] Stack : a800000802649400 0000000000000000 0000000000000000 a8000008139ffdd0 [ 34.196623] 0000000000000fba a800000808ee0000 0000000000000001 a8000008130c3e80 [ 34.204676] a8000008080d1280 a8000008139ffd58 a8000008139ffd58 1dbd2b22ea1dd500 [ 34.212729] a800000802649400 a800000808ee0000 ffffffffffffffea 0000000000000001 [ 34.220783] 0000000000001000 0000000000000000 00000001200ae518 ffffffffffffffff [ 34.228836] 000000fffbe0e530 a80000080837edf4 000000fffbe0e530 0000000000000000 [ 34.236890] 0000000000000000 0000000000000000 000000014ebb55a0 0000000000001000 [ 34.244943] 0000000000000001 a800000802649400 0000000000000000 0000000000000000 [ 34.252996] 0000000000000000 0000400400000000 0000000000000000 1dbd2b22ea1dd500 [ 34.261049] 00000000140000e3 a800000802649400 a800000802649400 a800000808ee0000 [ 34.269103] ... [ 34.271568] Call Trace: [ 34.274026] [<a80000080838e01c>] anon_pipe_read+0xd4/0x428 [ 34.279533] [<a80000080837edf4>] vfs_read+0x25c/0x318 [ 34.284607] [<a80000080837faac>] ksys_read+0x104/0x138 [ 34.289763] [<a80000080802b9cc>] syscall_common+0x44/0x68 [ 34.295187] [ 34.296689] Code: f84000cf 02209825 de020010 <dc420000> d8400004 02002825 0040f809 02802025 f84000c3 [ 34.306504] [ 34.308099] ---[ end trace 0000000000000000 ]--- My initial reproducer was the xdp-tools test suite. A standalone reproducer would be an lld/scd loop that, when the read is reordered by the CPU, triggers a fault. We can achieve this from userspace by stressing an anonymous pipe, which uses a mutex. Program used: // SPDX-License-Identifier: GPL-2.0 // pipe_race.c - reproducer for MIPS LL/SC reordering vs fs/pipe.c // // Two userspace processes on an anonymous pipe: // parent = writer: tight write() loop // child = reader: tight read() loop #define _GNU_SOURCE #include <assert.h> #include <errno.h> #include <sched.h> #include <signal.h> #include <stdio.h> #include <stdlib.h> #include <string.h> #include <sys/types.h> #include ---truncated---

drm/msm: RCU-free the scheduler-containing ring and VM objects

In the Linux kernel, the following vulnerability has been resolved: drm/msm: RCU-free the scheduler-containing ring and VM objects Both struct msm_ringbuffer and struct msm_gem_vm embed a struct drm_gpu_scheduler. msm_ringbuffer_destroy() and the VM free callback msm_gem_vm_free() call drm_sched_fini() on the embedded scheduler and then free the containing object with plain kfree(). drm_sched_fence_get_timeline_name() returns fence->sched->name, and the scheduler fence keeps a .release callback so it is not ops-detached on signalling. A finished fence exported to userspace (the submit out-fence, or a VM_BIND fence, via sync_file / drm_syncobj) keeps pointing at the embedded scheduler after the ring/VM is freed, so a later get_timeline_name() -- reachable unprivileged through SYNC_IOC_FILE_INFO -- dereferences freed slab memory (KASAN slab-use-after-free read). Per the dma-fence lifetime contract the exporter must keep the data backing a signalled fence alive for an RCU grace period. Free the scheduler-containing objects with kfree_rcu() instead of kfree(). Patchwork: https://patchwork.freedesktop.org/patch/750234/

drm/amdkfd: Avoid integer underflow with ffs in EOP ring size calc

In the Linux kernel, the following vulnerability has been resolved: drm/amdkfd: Avoid integer underflow with ffs in EOP ring size calc The low 6 bits of cp_hqd_eop_control store the base-2 logarithm of the EOP ring size. This was calculated as ffs(q->eop_ring_buffer_size / sizeof(unsigned int)) - 1 - 1 But ffs can in theory return 1 or 0, so this could underflow (although in practice the ring buffer size cannot be less than 4096). Change this to ffs(q->eop_ring_buffer_size / sizeof(unsigned int) / 4) using properties of logarithms. (cherry picked from commit 4f18c56630383c14bfc6b2d65f88f2f895d2121a)

Cleartext storage of sensitive information in the user account data in Devolutions Server 2026.3.12.0 and earlier allows an attacker with read access to the database or a database backup to obtain users' TOTP shared secrets via the unencrypted copy of the security key written when an administrator saves a user.

Server-side request forgery in the Active Directory credential validation in Devolutions Server 2026.3.12.0 and earlier allows an authenticated low-privileged user to capture the NetNTLMv2 hash of the DVLS service account via an attacker-controlled domain name in the request.

Improper release of expired privileged access checkouts in the PAM scheduled check-in task in Devolutions Server 2025.3.2.0 through 2026.2.22.0 allows an authenticated user with checkout rights on a PAM account to retain access to that account on the target system after the checkout expires via a backlog of 21 or more newer active checkouts that prevents the expired checkout from being checked in and its password from being rotated.

pgEdge RAG Server unbounded corpus read allows resource exhaustion from unauthenticated requests

pgEdge RAG Server is an API server for retrieval-augmented generation over PostgreSQL content using pgvector. Prior to 2.0.0, the unauthenticated query endpoint uses FetchDocuments in internal/database/search.go to read the entire filtered corpus without a LIMIT and rebuild an in-memory BM25 index on every request. Because HybridEnabled is set by default in internal/config/loader.go and the endpoint has no rate limit, inexpensive remote queries can impose database and CPU work proportional to corpus size and exhaust service resources. This issue is fixed in version 2.0.0.

In multiple functions of ipphelper.c, there is a possible stack bugger overflow due to a missing bounds check. This could lead to remote denial of service with no additional execution privileges needed. User interaction is not needed for exploitation.

In setAccessibilityWindowAttributes of AccessibilityWindowManager.java, there is a possible untrusted/misleading UI issue due to a missing permission check. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

Permission Bypass in Android WifiP2P Service Allows Local Privilege Escalation

In getMessenger of WifiP2pServiceImpl.java, there is a possible permission bypass due to a missing permission check. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

In set_type4_frame_classifier of sta_iface.cpp, there is a possible memory safety issue due to a heap buffer overflow. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

In descramble of multiple files, there is a possible out-of-bounds write due to improper casting. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.