Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/xe/migrate: prevent infinite recursion<br /> <br /> If the buf + offset is not aligned to XE_CAHELINE_BYTES we fallback to<br /> using a bounce buffer. However the bounce buffer here is allocated on<br /> the stack, and the only alignment requirement here is that it&amp;#39;s<br /> naturally aligned to u8, and not XE_CACHELINE_BYTES. If the bounce<br /> buffer is also misaligned we then recurse back into the function again,<br /> however the new bounce buffer might also not be aligned, and might never<br /> be until we eventually blow through the stack, as we keep recursing.<br /> <br /> Instead of using the stack use kmalloc, which should respect the<br /> power-of-two alignment request here. Fixes a kernel panic when<br /> triggering this path through eudebug.<br /> <br /> v2 (Stuart):<br /> - Add build bug check for power-of-two restriction<br /> - s/EINVAL/ENOMEM/<br /> <br /> (cherry picked from commit 38b34e928a08ba594c4bbf7118aa3aadacd62fff)

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> x86/fpu: Fix NULL dereference in avx512_status()<br /> <br /> Problem<br /> -------<br /> With CONFIG_X86_DEBUG_FPU enabled, reading /proc/[kthread]/arch_status<br /> causes a warning and a NULL pointer dereference.<br /> <br /> This is because the AVX-512 timestamp code uses x86_task_fpu() but<br /> doesn&amp;#39;t check it for NULL. CONFIG_X86_DEBUG_FPU addles that function<br /> for kernel threads (PF_KTHREAD specifically), making it return NULL.<br /> <br /> The point of the warning was to ensure that kernel threads only access<br /> task-&gt;fpu after going through kernel_fpu_begin()/_end(). Note: all<br /> kernel tasks exposed in /proc have a valid task-&gt;fpu.<br /> <br /> Solution<br /> --------<br /> One option is to silence the warning and check for NULL from<br /> x86_task_fpu(). However, that warning is fairly fresh and seems like a<br /> defense against misuse of the FPU state in kernel threads.<br /> <br /> Instead, stop outputting AVX-512_elapsed_ms for kernel threads<br /> altogether. The data was garbage anyway because avx512_timestamp is<br /> only updated for user threads, not kernel threads.<br /> <br /> If anyone ever wants to track kernel thread AVX-512 use, they can come<br /> back later and do it properly, separate from this bug fix.<br /> <br /> [ dhansen: mostly rewrite changelog ]

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fbdev: Fix vmalloc out-of-bounds write in fast_imageblit<br /> <br /> This issue triggers when a userspace program does an ioctl<br /> FBIOPUT_CON2FBMAP by passing console number and frame buffer number.<br /> Ideally this maps console to frame buffer and updates the screen if<br /> console is visible.<br /> <br /> As part of mapping it has to do resize of console according to frame<br /> buffer info. if this resize fails and returns from vc_do_resize() and<br /> continues further. At this point console and new frame buffer are mapped<br /> and sets display vars. Despite failure still it continue to proceed<br /> updating the screen at later stages where vc_data is related to previous<br /> frame buffer and frame buffer info and display vars are mapped to new<br /> frame buffer and eventully leading to out-of-bounds write in<br /> fast_imageblit(). This bheviour is excepted only when fg_console is<br /> equal to requested console which is a visible console and updates screen<br /> with invalid struct references in fbcon_putcs().

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net/sched: ets: use old &amp;#39;nbands&amp;#39; while purging unused classes<br /> <br /> Shuang reported sch_ets test-case [1] crashing in ets_class_qlen_notify()<br /> after recent changes from Lion [2]. The problem is: in ets_qdisc_change()<br /> we purge unused DWRR queues; the value of &amp;#39;q-&gt;nbands&amp;#39; is the new one, and<br /> the cleanup should be done with the old one. The problem is here since my<br /> first attempts to fix ets_qdisc_change(), but it surfaced again after the<br /> recent qdisc len accounting fixes. Fix it purging idle DWRR queues before<br /> assigning a new value of &amp;#39;q-&gt;nbands&amp;#39;, so that all purge operations find a<br /> consistent configuration:<br /> <br /> - old &amp;#39;q-&gt;nbands&amp;#39; because it&amp;#39;s needed by ets_class_find()<br /> - old &amp;#39;q-&gt;nstrict&amp;#39; because it&amp;#39;s needed by ets_class_is_strict()<br /> <br /> BUG: kernel NULL pointer dereference, address: 0000000000000000<br /> #PF: supervisor read access in kernel mode<br /> #PF: error_code(0x0000) - not-present page<br /> PGD 0 P4D 0<br /> Oops: Oops: 0000 [#1] SMP NOPTI<br /> CPU: 62 UID: 0 PID: 39457 Comm: tc Kdump: loaded Not tainted 6.12.0-116.el10.x86_64 #1 PREEMPT(voluntary)<br /> Hardware name: Dell Inc. PowerEdge R640/06DKY5, BIOS 2.12.2 07/09/2021<br /> RIP: 0010:__list_del_entry_valid_or_report+0x4/0x80<br /> Code: ff 4c 39 c7 0f 84 39 19 8e ff b8 01 00 00 00 c3 cc cc cc cc 66 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 f3 0f 1e fa 8b 17 48 8b 4f 08 48 85 d2 0f 84 56 19 8e ff 48 85 c9 0f 84 ab<br /> RSP: 0018:ffffba186009f400 EFLAGS: 00010202<br /> RAX: 00000000000000d6 RBX: 0000000000000000 RCX: 0000000000000004<br /> RDX: ffff9f0fa29b69c0 RSI: 0000000000000000 RDI: 0000000000000000<br /> RBP: ffffffffc12c2400 R08: 0000000000000008 R09: 0000000000000004<br /> R10: ffffffffffffffff R11: 0000000000000004 R12: 0000000000000000<br /> R13: ffff9f0f8cfe0000 R14: 0000000000100005 R15: 0000000000000000<br /> FS: 00007f2154f37480(0000) GS:ffff9f269c1c0000(0000) knlGS:0000000000000000<br /> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> CR2: 0000000000000000 CR3: 00000001530be001 CR4: 00000000007726f0<br /> DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000<br /> DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400<br /> PKRU: 55555554<br /> Call Trace:<br /> <br /> ets_class_qlen_notify+0x65/0x90 [sch_ets]<br /> qdisc_tree_reduce_backlog+0x74/0x110<br /> ets_qdisc_change+0x630/0xa40 [sch_ets]<br /> __tc_modify_qdisc.constprop.0+0x216/0x7f0<br /> tc_modify_qdisc+0x7c/0x120<br /> rtnetlink_rcv_msg+0x145/0x3f0<br /> netlink_rcv_skb+0x53/0x100<br /> netlink_unicast+0x245/0x390<br /> netlink_sendmsg+0x21b/0x470<br /> ____sys_sendmsg+0x39d/0x3d0<br /> ___sys_sendmsg+0x9a/0xe0<br /> __sys_sendmsg+0x7a/0xd0<br /> do_syscall_64+0x7d/0x160<br /> entry_SYSCALL_64_after_hwframe+0x76/0x7e<br /> RIP: 0033:0x7f2155114084<br /> Code: 89 02 b8 ff ff ff ff eb bb 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 f3 0f 1e fa 80 3d 25 f0 0c 00 00 74 13 b8 2e 00 00 00 0f 05 3d 00 f0 ff ff 77 54 c3 0f 1f 00 48 83 ec 28 89 54 24 1c 48 89<br /> RSP: 002b:00007fff1fd7a988 EFLAGS: 00000202 ORIG_RAX: 000000000000002e<br /> RAX: ffffffffffffffda RBX: 0000560ec063e5e0 RCX: 00007f2155114084<br /> RDX: 0000000000000000 RSI: 00007fff1fd7a9f0 RDI: 0000000000000003<br /> RBP: 00007fff1fd7aa60 R08: 0000000000000010 R09: 000000000000003f<br /> R10: 0000560ee9b3a010 R11: 0000000000000202 R12: 00007fff1fd7aae0<br /> R13: 000000006891ccde R14: 0000560ec063e5e0 R15: 00007fff1fd7aad0<br /> <br /> <br /> [1] https://lore.kernel.org/netdev/e08c7f4a6882f260011909a868311c6e9b54f3e4.1639153474.git.dcaratti@redhat.com/<br /> [2] https://lore.kernel.org/netdev/d912cbd7-193b-4269-9857-525bee8bbb6a@gmail.com/

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> hv_netvsc: Fix panic during namespace deletion with VF<br /> <br /> The existing code move the VF NIC to new namespace when NETDEV_REGISTER is<br /> received on netvsc NIC. During deletion of the namespace,<br /> default_device_exit_batch() &gt;&gt; default_device_exit_net() is called. When<br /> netvsc NIC is moved back and registered to the default namespace, it<br /> automatically brings VF NIC back to the default namespace. This will cause<br /> the default_device_exit_net() &gt;&gt; for_each_netdev_safe loop unable to detect<br /> the list end, and hit NULL ptr:<br /> <br /> [ 231.449420] mana 7870:00:00.0 enP30832s1: Moved VF to namespace with: eth0<br /> [ 231.449656] BUG: kernel NULL pointer dereference, address: 0000000000000010<br /> [ 231.450246] #PF: supervisor read access in kernel mode<br /> [ 231.450579] #PF: error_code(0x0000) - not-present page<br /> [ 231.450916] PGD 17b8a8067 P4D 0<br /> [ 231.451163] Oops: Oops: 0000 [#1] SMP NOPTI<br /> [ 231.451450] CPU: 82 UID: 0 PID: 1394 Comm: kworker/u768:1 Not tainted 6.16.0-rc4+ #3 VOLUNTARY<br /> [ 231.452042] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 11/21/2024<br /> [ 231.452692] Workqueue: netns cleanup_net<br /> [ 231.452947] RIP: 0010:default_device_exit_batch+0x16c/0x3f0<br /> [ 231.453326] Code: c0 0c f5 b3 e8 d5 db fe ff 48 85 c0 74 15 48 c7 c2 f8 fd ca b2 be 10 00 00 00 48 8d 7d c0 e8 7b 77 25 00 49 8b 86 28 01 00 00 8b 50 10 4c 8b 2a 4c 8d 62 f0 49 83 ed 10 4c 39 e0 0f 84 d6 00<br /> [ 231.454294] RSP: 0018:ff75fc7c9bf9fd00 EFLAGS: 00010246<br /> [ 231.454610] RAX: 0000000000000000 RBX: 0000000000000002 RCX: 61c8864680b583eb<br /> [ 231.455094] RDX: ff1fa9f71462d800 RSI: ff75fc7c9bf9fd38 RDI: 0000000030766564<br /> [ 231.455686] RBP: ff75fc7c9bf9fd78 R08: 0000000000000000 R09: 0000000000000000<br /> [ 231.456126] R10: 0000000000000001 R11: 0000000000000004 R12: ff1fa9f70088e340<br /> [ 231.456621] R13: ff1fa9f70088e340 R14: ffffffffb3f50c20 R15: ff1fa9f7103e6340<br /> [ 231.457161] FS: 0000000000000000(0000) GS:ff1faa6783a08000(0000) knlGS:0000000000000000<br /> [ 231.457707] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033<br /> [ 231.458031] CR2: 0000000000000010 CR3: 0000000179ab2006 CR4: 0000000000b73ef0<br /> [ 231.458434] Call Trace:<br /> [ 231.458600] <br /> [ 231.458777] ops_undo_list+0x100/0x220<br /> [ 231.459015] cleanup_net+0x1b8/0x300<br /> [ 231.459285] process_one_work+0x184/0x340<br /> <br /> To fix it, move the ns change to a workqueue, and take rtnl_lock to avoid<br /> changing the netdev list when default_device_exit_net() is using it.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> comedi: fix race between polling and detaching<br /> <br /> syzbot reports a use-after-free in comedi in the below link, which is<br /> due to comedi gladly removing the allocated async area even though poll<br /> requests are still active on the wait_queue_head inside of it. This can<br /> cause a use-after-free when the poll entries are later triggered or<br /> removed, as the memory for the wait_queue_head has been freed. We need<br /> to check there are no tasks queued on any of the subdevices&amp;#39; wait queues<br /> before allowing the device to be detached by the `COMEDI_DEVCONFIG`<br /> ioctl.<br /> <br /> Tasks will read-lock `dev-&gt;attach_lock` before adding themselves to the<br /> subdevice wait queue, so fix the problem in the `COMEDI_DEVCONFIG` ioctl<br /> handler by write-locking `dev-&gt;attach_lock` before checking that all of<br /> the subdevices are safe to be deleted. This includes testing for any<br /> sleepers on the subdevices&amp;#39; wait queues. It remains locked until the<br /> device has been detached. This requires the `comedi_device_detach()`<br /> function to be refactored slightly, moving the bulk of it into new<br /> function `comedi_device_detach_locked()`.<br /> <br /> Note that the refactor of `comedi_device_detach()` results in<br /> `comedi_device_cancel_all()` now being called while `dev-&gt;attach_lock`<br /> is write-locked, which wasn&amp;#39;t the case previously, but that does not<br /> matter.<br /> <br /> Thanks to Jens Axboe for diagnosing the problem and co-developing this<br /> patch.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> iommufd: Prevent ALIGN() overflow<br /> <br /> When allocating IOVA the candidate range gets aligned to the target<br /> alignment. If the range is close to ULONG_MAX then the ALIGN() can<br /> wrap resulting in a corrupted iova.<br /> <br /> Open code the ALIGN() using get_add_overflow() to prevent this.<br /> This simplifies the checks as we don&amp;#39;t need to check for length earlier<br /> either.<br /> <br /> Consolidate the two copies of this code under a single helper.<br /> <br /> This bug would allow userspace to create a mapping that overlaps with some<br /> other mapping or a reserved range.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> userfaultfd: fix a crash in UFFDIO_MOVE when PMD is a migration entry<br /> <br /> When UFFDIO_MOVE encounters a migration PMD entry, it proceeds with<br /> obtaining a folio and accessing it even though the entry is swp_entry_t. <br /> Add the missing check and let split_huge_pmd() handle migration entries. <br /> While at it also remove unnecessary folio check.<br /> <br /> [surenb@google.com: remove extra folio check, per David]

NVIDIA HGX and DGX contain a vulnerability where a misconfiguration of the LS10 could enable an attacker to set an unsafe debug access level. A successful exploit of this vulnerability might lead to denial of service.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> media: venus: Fix OOB read due to missing payload bound check<br /> <br /> Currently, The event_seq_changed() handler processes a variable number<br /> of properties sent by the firmware. The number of properties is indicated<br /> by the firmware and used to iterate over the payload. However, the<br /> payload size is not being validated against the actual message length.<br /> <br /> This can lead to out-of-bounds memory access if the firmware provides a<br /> property count that exceeds the data available in the payload. Such a<br /> condition can result in kernel crashes or potential information leaks if<br /> memory beyond the buffer is accessed.<br /> <br /> Fix this by properly validating the remaining size of the payload before<br /> each property access and updating bounds accordingly as properties are<br /> parsed.<br /> <br /> This ensures that property parsing is safely bounded within the received<br /> message buffer and protects against malformed or malicious firmware<br /> behavior.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> mm/ptdump: take the memory hotplug lock inside ptdump_walk_pgd()<br /> <br /> Memory hot remove unmaps and tears down various kernel page table regions<br /> as required. The ptdump code can race with concurrent modifications of<br /> the kernel page tables. When leaf entries are modified concurrently, the<br /> dump code may log stale or inconsistent information for a VA range, but<br /> this is otherwise not harmful.<br /> <br /> But when intermediate levels of kernel page table are freed, the dump code<br /> will continue to use memory that has been freed and potentially<br /> reallocated for another purpose. In such cases, the ptdump code may<br /> dereference bogus addresses, leading to a number of potential problems.<br /> <br /> To avoid the above mentioned race condition, platforms such as arm64,<br /> riscv and s390 take memory hotplug lock, while dumping kernel page table<br /> via the sysfs interface /sys/kernel/debug/kernel_page_tables.<br /> <br /> Similar race condition exists while checking for pages that might have<br /> been marked W+X via /sys/kernel/debug/kernel_page_tables/check_wx_pages<br /> which in turn calls ptdump_check_wx(). Instead of solving this race<br /> condition again, let&amp;#39;s just move the memory hotplug lock inside generic<br /> ptdump_check_wx() which will benefit both the scenarios.<br /> <br /> Drop get_online_mems() and put_online_mems() combination from all existing<br /> platform ptdump code paths.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> media: uvcvideo: Fix 1-byte out-of-bounds read in uvc_parse_format()<br /> <br /> The buffer length check before calling uvc_parse_format() only ensured<br /> that the buffer has at least 3 bytes (buflen &gt; 2), buf the function<br /> accesses buffer[3], requiring at least 4 bytes.<br /> <br /> This can lead to an out-of-bounds read if the buffer has exactly 3 bytes.<br /> <br /> Fix it by checking that the buffer has at least 4 bytes in<br /> uvc_parse_format().