Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net/mlx5: Fix double unregister of HCA_PORTS component<br /> <br /> Clear hca_devcom_comp in device&amp;#39;s private data after unregistering it in<br /> LAG teardown. Otherwise a slightly lagging second pass through<br /> mlx5_unload_one() might try to unregister it again and trip over<br /> use-after-free.<br /> <br /> On s390 almost all PCI level recovery events trigger two passes through<br /> mxl5_unload_one() - one through the poll_health() method and one through<br /> mlx5_pci_err_detected() as callback from generic PCI error recovery.<br /> While testing PCI error recovery paths with more kernel debug features<br /> enabled, this issue reproducibly led to kernel panics with the following<br /> call chain:<br /> <br /> Unable to handle kernel pointer dereference in virtual kernel address space<br /> Failing address: 6b6b6b6b6b6b6000 TEID: 6b6b6b6b6b6b6803 ESOP-2 FSI<br /> Fault in home space mode while using kernel ASCE.<br /> AS:00000000705c4007 R3:0000000000000024<br /> Oops: 0038 ilc:3 [#1]SMP<br /> <br /> CPU: 14 UID: 0 PID: 156 Comm: kmcheck Kdump: loaded Not tainted<br /> 6.18.0-20251130.rc7.git0.16131a59cab1.300.fc43.s390x+debug #1 PREEMPT<br /> <br /> Krnl PSW : 0404e00180000000 0000020fc86aa1dc (__lock_acquire+0x5c/0x15f0)<br /> R:0 T:1 IO:0 EX:0 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3<br /> Krnl GPRS: 0000000000000000 0000020f00000001 6b6b6b6b6b6b6c33 0000000000000000<br /> 0000000000000000 0000000000000000 0000000000000001 0000000000000000<br /> 0000000000000000 0000020fca28b820 0000000000000000 0000010a1ced8100<br /> 0000010a1ced8100 0000020fc9775068 0000018fce14f8b8 0000018fce14f7f8<br /> Krnl Code: 0000020fc86aa1cc: e3b003400004 lg %r11,832<br /> 0000020fc86aa1d2: a7840211 brc 8,0000020fc86aa5f4<br /> *0000020fc86aa1d6: c09000df0b25 larl %r9,0000020fca28b820<br /> &gt;0000020fc86aa1dc: d50790002000 clc 0(8,%r9),0(%r2)<br /> 0000020fc86aa1e2: a7840209 brc 8,0000020fc86aa5f4<br /> 0000020fc86aa1e6: c0e001100401 larl %r14,0000020fca8aa9e8<br /> 0000020fc86aa1ec: c01000e25a00 larl %r1,0000020fca2f55ec<br /> 0000020fc86aa1f2: a7eb00e8 aghi %r14,232<br /> <br /> Call Trace:<br /> __lock_acquire+0x5c/0x15f0<br /> lock_acquire.part.0+0xf8/0x270<br /> lock_acquire+0xb0/0x1b0<br /> down_write+0x5a/0x250<br /> mlx5_detach_device+0x42/0x110 [mlx5_core]<br /> mlx5_unload_one_devl_locked+0x50/0xc0 [mlx5_core]<br /> mlx5_unload_one+0x42/0x60 [mlx5_core]<br /> mlx5_pci_err_detected+0x94/0x150 [mlx5_core]<br /> zpci_event_attempt_error_recovery+0xcc/0x388

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fuse: missing copy_finish in fuse-over-io-uring argument copies<br /> <br /> Fix a possible reference count leak of payload pages during<br /> fuse argument copies.<br /> <br /> [Joanne: simplified error cleanup]

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> xfs: fix a UAF problem in xattr repair<br /> <br /> The xchk_setup_xattr_buf function can allocate a new value buffer, which<br /> means that any reference to ab-&gt;value before the call could become a<br /> dangling pointer. Fix this by moving an assignment to after the buffer<br /> setup.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ksmbd: skip lock-range check on equal size to avoid size==0 underflow<br /> <br /> When size equals the current i_size (including 0), the code used to call<br /> check_lock_range(filp, i_size, size - 1, WRITE), which computes `size - 1`<br /> and can underflow for size==0. Skip the equal case.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ALSA: usb-mixer: us16x08: validate meter packet indices<br /> <br /> get_meter_levels_from_urb() parses the 64-byte meter packets sent by<br /> the device and fills the per-channel arrays meter_level[],<br /> comp_level[] and master_level[] in struct snd_us16x08_meter_store.<br /> <br /> Currently the function derives the channel index directly from the<br /> meter packet (MUB2(meter_urb, s) - 1) and uses it to index those<br /> arrays without validating the range. If the packet contains a<br /> negative or out-of-range channel number, the driver may write past<br /> the end of these arrays.<br /> <br /> Introduce a local channel variable and validate it before updating the<br /> arrays. We reject negative indices, limit meter_level[] and<br /> comp_level[] to SND_US16X08_MAX_CHANNELS, and guard master_level[]<br /> updates with ARRAY_SIZE(master_level).

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net: openvswitch: fix middle attribute validation in push_nsh() action<br /> <br /> The push_nsh() action structure looks like this:<br /> <br /> OVS_ACTION_ATTR_PUSH_NSH(OVS_KEY_ATTR_NSH(OVS_NSH_KEY_ATTR_BASE,...))<br /> <br /> The outermost OVS_ACTION_ATTR_PUSH_NSH attribute is OK&amp;#39;ed by the<br /> nla_for_each_nested() inside __ovs_nla_copy_actions(). The innermost<br /> OVS_NSH_KEY_ATTR_BASE/MD1/MD2 are OK&amp;#39;ed by the nla_for_each_nested()<br /> inside nsh_key_put_from_nlattr(). But nothing checks if the attribute<br /> in the middle is OK. We don&amp;#39;t even check that this attribute is the<br /> OVS_KEY_ATTR_NSH. We just do a double unwrap with a pair of nla_data()<br /> calls - first time directly while calling validate_push_nsh() and the<br /> second time as part of the nla_for_each_nested() macro, which isn&amp;#39;t<br /> safe, potentially causing invalid memory access if the size of this<br /> attribute is incorrect. The failure may not be noticed during<br /> validation due to larger netlink buffer, but cause trouble later during<br /> action execution where the buffer is allocated exactly to the size:<br /> <br /> BUG: KASAN: slab-out-of-bounds in nsh_hdr_from_nlattr+0x1dd/0x6a0 [openvswitch]<br /> Read of size 184 at addr ffff88816459a634 by task a.out/22624<br /> <br /> CPU: 8 UID: 0 PID: 22624 6.18.0-rc7+ #115 PREEMPT(voluntary)<br /> Call Trace:<br /> <br /> dump_stack_lvl+0x51/0x70<br /> print_address_description.constprop.0+0x2c/0x390<br /> kasan_report+0xdd/0x110<br /> kasan_check_range+0x35/0x1b0<br /> __asan_memcpy+0x20/0x60<br /> nsh_hdr_from_nlattr+0x1dd/0x6a0 [openvswitch]<br /> push_nsh+0x82/0x120 [openvswitch]<br /> do_execute_actions+0x1405/0x2840 [openvswitch]<br /> ovs_execute_actions+0xd5/0x3b0 [openvswitch]<br /> ovs_packet_cmd_execute+0x949/0xdb0 [openvswitch]<br /> genl_family_rcv_msg_doit+0x1d6/0x2b0<br /> genl_family_rcv_msg+0x336/0x580<br /> genl_rcv_msg+0x9f/0x130<br /> netlink_rcv_skb+0x11f/0x370<br /> genl_rcv+0x24/0x40<br /> netlink_unicast+0x73e/0xaa0<br /> netlink_sendmsg+0x744/0xbf0<br /> __sys_sendto+0x3d6/0x450<br /> do_syscall_64+0x79/0x2c0<br /> entry_SYSCALL_64_after_hwframe+0x76/0x7e<br /> <br /> <br /> Let&amp;#39;s add some checks that the attribute is properly sized and it&amp;#39;s<br /> the only one attribute inside the action. Technically, there is no<br /> real reason for OVS_KEY_ATTR_NSH to be there, as we know that we&amp;#39;re<br /> pushing an NSH header already, it just creates extra nesting, but<br /> that&amp;#39;s how uAPI works today. So, keeping as it is.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> netrom: Fix memory leak in nr_sendmsg()<br /> <br /> syzbot reported a memory leak [1].<br /> <br /> When function sock_alloc_send_skb() return NULL in nr_output(), the<br /> original skb is not freed, which was allocated in nr_sendmsg(). Fix this<br /> by freeing it before return.<br /> <br /> [1]<br /> BUG: memory leak<br /> unreferenced object 0xffff888129f35500 (size 240):<br /> comm "syz.0.17", pid 6119, jiffies 4294944652<br /> hex dump (first 32 bytes):<br /> 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................<br /> 00 00 00 00 00 00 00 00 00 10 52 28 81 88 ff ff ..........R(....<br /> backtrace (crc 1456a3e4):<br /> kmemleak_alloc_recursive include/linux/kmemleak.h:44 [inline]<br /> slab_post_alloc_hook mm/slub.c:4983 [inline]<br /> slab_alloc_node mm/slub.c:5288 [inline]<br /> kmem_cache_alloc_node_noprof+0x36f/0x5e0 mm/slub.c:5340<br /> __alloc_skb+0x203/0x240 net/core/skbuff.c:660<br /> alloc_skb include/linux/skbuff.h:1383 [inline]<br /> alloc_skb_with_frags+0x69/0x3f0 net/core/skbuff.c:6671<br /> sock_alloc_send_pskb+0x379/0x3e0 net/core/sock.c:2965<br /> sock_alloc_send_skb include/net/sock.h:1859 [inline]<br /> nr_sendmsg+0x287/0x450 net/netrom/af_netrom.c:1105<br /> sock_sendmsg_nosec net/socket.c:727 [inline]<br /> __sock_sendmsg net/socket.c:742 [inline]<br /> sock_write_iter+0x293/0x2a0 net/socket.c:1195<br /> new_sync_write fs/read_write.c:593 [inline]<br /> vfs_write+0x45d/0x710 fs/read_write.c:686<br /> ksys_write+0x143/0x170 fs/read_write.c:738<br /> do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]<br /> do_syscall_64+0xa4/0xfa0 arch/x86/entry/syscall_64.c:94<br /> entry_SYSCALL_64_after_hwframe+0x77/0x7f

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fsnotify: do not generate ACCESS/MODIFY events on child for special files<br /> <br /> inotify/fanotify do not allow users with no read access to a file to<br /> subscribe to events (e.g. IN_ACCESS/IN_MODIFY), but they do allow the<br /> same user to subscribe for watching events on children when the user<br /> has access to the parent directory (e.g. /dev).<br /> <br /> Users with no read access to a file but with read access to its parent<br /> directory can still stat the file and see if it was accessed/modified<br /> via atime/mtime change.<br /> <br /> The same is not true for special files (e.g. /dev/null). Users will not<br /> generally observe atime/mtime changes when other users read/write to<br /> special files, only when someone sets atime/mtime via utimensat().<br /> <br /> Align fsnotify events with this stat behavior and do not generate<br /> ACCESS/MODIFY events to parent watchers on read/write of special files.<br /> The events are still generated to parent watchers on utimensat(). This<br /> closes some side-channels that could be possibly used for information<br /> exfiltration [1].<br /> <br /> [1] https://snee.la/pdf/pubs/file-notification-attacks.pdf

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> hwmon: (ibmpex) fix use-after-free in high/low store<br /> <br /> The ibmpex_high_low_store() function retrieves driver data using<br /> dev_get_drvdata() and uses it without validation. This creates a race<br /> condition where the sysfs callback can be invoked after the data<br /> structure is freed, leading to use-after-free.<br /> <br /> Fix by adding a NULL check after dev_get_drvdata(), and reordering<br /> operations in the deletion path to prevent TOCTOU.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net/handshake: duplicate handshake cancellations leak socket<br /> <br /> When a handshake request is cancelled it is removed from the<br /> handshake_net-&gt;hn_requests list, but it is still present in the<br /> handshake_rhashtbl until it is destroyed.<br /> <br /> If a second cancellation request arrives for the same handshake request,<br /> then remove_pending() will return false... and assuming<br /> HANDSHAKE_F_REQ_COMPLETED isn&amp;#39;t set in req-&gt;hr_flags, we&amp;#39;ll continue<br /> processing through the out_true label, where we put another reference on<br /> the sock and a refcount underflow occurs.<br /> <br /> This can happen for example if a handshake times out - particularly if<br /> the SUNRPC client sends the AUTH_TLS probe to the server but doesn&amp;#39;t<br /> follow it up with the ClientHello due to a problem with tlshd. When the<br /> timeout is hit on the server, the server will send a FIN, which triggers<br /> a cancellation request via xs_reset_transport(). When the timeout is<br /> hit on the client, another cancellation request happens via<br /> xs_tls_handshake_sync().<br /> <br /> Add a test_and_set_bit(HANDSHAKE_F_REQ_COMPLETED) in the pending cancel<br /> path so duplicate cancels can be detected.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> btrfs: don&amp;#39;t log conflicting inode if it&amp;#39;s a dir moved in the current transaction<br /> <br /> We can&amp;#39;t log a conflicting inode if it&amp;#39;s a directory and it was moved<br /> from one parent directory to another parent directory in the current<br /> transaction, as this can result an attempt to have a directory with<br /> two hard links during log replay, one for the old parent directory and<br /> another for the new parent directory.<br /> <br /> The following scenario triggers that issue:<br /> <br /> 1) We have directories "dir1" and "dir2" created in a past transaction.<br /> Directory "dir1" has inode A as its parent directory;<br /> <br /> 2) We move "dir1" to some other directory;<br /> <br /> 3) We create a file with the name "dir1" in directory inode A;<br /> <br /> 4) We fsync the new file. This results in logging the inode of the new file<br /> and the inode for the directory "dir1" that was previously moved in the<br /> current transaction. So the log tree has the INODE_REF item for the<br /> new location of "dir1";<br /> <br /> 5) We move the new file to some other directory. This results in updating<br /> the log tree to included the new INODE_REF for the new location of the<br /> file and removes the INODE_REF for the old location. This happens<br /> during the rename when we call btrfs_log_new_name();<br /> <br /> 6) We fsync the file, and that persists the log tree changes done in the<br /> previous step (btrfs_log_new_name() only updates the log tree in<br /> memory);<br /> <br /> 7) We have a power failure;<br /> <br /> 8) Next time the fs is mounted, log replay happens and when processing<br /> the inode for directory "dir1" we find a new INODE_REF and add that<br /> link, but we don&amp;#39;t remove the old link of the inode since we have<br /> not logged the old parent directory of the directory inode "dir1".<br /> <br /> As a result after log replay finishes when we trigger writeback of the<br /> subvolume tree&amp;#39;s extent buffers, the tree check will detect that we have<br /> a directory a hard link count of 2 and we get a mount failure.<br /> The errors and stack traces reported in dmesg/syslog are like this:<br /> <br /> [ 3845.729764] BTRFS info (device dm-0): start tree-log replay<br /> [ 3845.730304] page: refcount:3 mapcount:0 mapping:000000005c8a3027 index:0x1d00 pfn:0x11510c<br /> [ 3845.731236] memcg:ffff9264c02f4e00<br /> [ 3845.731751] aops:btree_aops [btrfs] ino:1<br /> [ 3845.732300] flags: 0x17fffc00000400a(uptodate|private|writeback|node=0|zone=2|lastcpupid=0x1ffff)<br /> [ 3845.733346] raw: 017fffc00000400a 0000000000000000 dead000000000122 ffff9264d978aea8<br /> [ 3845.734265] raw: 0000000000001d00 ffff92650e6d4738 00000003ffffffff ffff9264c02f4e00<br /> [ 3845.735305] page dumped because: eb page dump<br /> [ 3845.735981] BTRFS critical (device dm-0): corrupt leaf: root=5 block=30408704 slot=6 ino=257, invalid nlink: has 2 expect no more than 1 for dir<br /> [ 3845.737786] BTRFS info (device dm-0): leaf 30408704 gen 10 total ptrs 17 free space 14881 owner 5<br /> [ 3845.737789] BTRFS info (device dm-0): refs 4 lock_owner 0 current 30701<br /> [ 3845.737792] item 0 key (256 INODE_ITEM 0) itemoff 16123 itemsize 160<br /> [ 3845.737794] inode generation 3 transid 9 size 16 nbytes 16384<br /> [ 3845.737795] block group 0 mode 40755 links 1 uid 0 gid 0<br /> [ 3845.737797] rdev 0 sequence 2 flags 0x0<br /> [ 3845.737798] atime 1764259517.0<br /> [ 3845.737800] ctime 1764259517.572889464<br /> [ 3845.737801] mtime 1764259517.572889464<br /> [ 3845.737802] otime 1764259517.0<br /> [ 3845.737803] item 1 key (256 INODE_REF 256) itemoff 16111 itemsize 12<br /> [ 3845.737805] index 0 name_len 2<br /> [ 3845.737807] item 2 key (256 DIR_ITEM 2363071922) itemoff 16077 itemsize 34<br /> [ 3845.737808] location key (257 1 0) type 2<br /> [ 3845.737810] transid 9 data_len 0 name_len 4<br /> [ 3845.737811] item 3 key (256 DIR_ITEM 2676584006) itemoff 16043 itemsize 34<br /> [ 3845.737813] location key (258 1 0) type 2<br /> [ 3845.737814] transid 9 data_len 0 name_len 4<br /> [ 3845.737815] item 4 key (256 DIR_INDEX 2) itemoff 16009 itemsize 34<br /> [ 3845.737816] location key (257 1 0) type 2<br /> [<br /> ---truncated---

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net/mlx5e: Avoid unregistering PSP twice<br /> <br /> PSP is unregistered twice in:<br /> _mlx5e_remove -&gt; mlx5e_psp_unregister<br /> mlx5e_nic_cleanup -&gt; mlx5e_psp_unregister<br /> <br /> This leads to a refcount underflow in some conditions:<br /> ------------[ cut here ]------------<br /> refcount_t: underflow; use-after-free.<br /> WARNING: CPU: 2 PID: 1694 at lib/refcount.c:28 refcount_warn_saturate+0xd8/0xe0<br /> [...]<br /> mlx5e_psp_unregister+0x26/0x50 [mlx5_core]<br /> mlx5e_nic_cleanup+0x26/0x90 [mlx5_core]<br /> mlx5e_remove+0xe6/0x1f0 [mlx5_core]<br /> auxiliary_bus_remove+0x18/0x30<br /> device_release_driver_internal+0x194/0x1f0<br /> bus_remove_device+0xc6/0x130<br /> device_del+0x159/0x3c0<br /> mlx5_rescan_drivers_locked+0xbc/0x2a0 [mlx5_core]<br /> [...]<br /> <br /> Do not directly remove psp from the _mlx5e_remove path, the PSP cleanup<br /> happens as part of profile cleanup.