Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> fbdev: Fix invalid page access after closing deferred I/O devices<br /> <br /> When a fbdev with deferred I/O is once opened and closed, the dirty<br /> pages still remain queued in the pageref list, and eventually later<br /> those may be processed in the delayed work. This may lead to a<br /> corruption of pages, hitting an Oops.<br /> <br /> This patch makes sure to cancel the delayed work and clean up the<br /> pageref list at closing the device for addressing the bug. A part of<br /> the cleanup code is factored out as a new helper function that is<br /> called from the common fb_release().

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ceph: blocklist the kclient when receiving corrupted snap trace<br /> <br /> When received corrupted snap trace we don&amp;#39;t know what exactly has<br /> happened in MDS side. And we shouldn&amp;#39;t continue IOs and metadatas<br /> access to MDS, which may corrupt or get incorrect contents.<br /> <br /> This patch will just block all the further IO/MDS requests<br /> immediately and then evict the kclient itself.<br /> <br /> The reason why we still need to evict the kclient just after<br /> blocking all the further IOs is that the MDS could revoke the caps<br /> faster.

Rejected reason: This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

Rejected reason: This CVE ID has been rejected or withdrawn by its CVE Numbering Authority.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> bpf, sockmap: Don&amp;#39;t let sock_map_{close,destroy,unhash} call itself<br /> <br /> sock_map proto callbacks should never call themselves by design. Protect<br /> against bugs like [1] and break out of the recursive loop to avoid a stack<br /> overflow in favor of a resource leak.<br /> <br /> [1] https://lore.kernel.org/all/00000000000073b14905ef2e7401@google.com/

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> ALSA: hda: Do not unset preset when cleaning up codec<br /> <br /> Several functions that take part in codec&amp;#39;s initialization and removal<br /> are re-used by ASoC codec drivers implementations. Drivers mimic the<br /> behavior of hda_codec_driver_probe/remove() found in<br /> sound/pci/hda/hda_bind.c with their component-&gt;probe/remove() instead.<br /> <br /> One of the reasons for that is the expectation of<br /> snd_hda_codec_device_new() to receive a valid pointer to an instance of<br /> struct snd_card. This expectation can be met only once sound card<br /> components probing commences.<br /> <br /> As ASoC sound card may be unbound without codec device being actually<br /> removed from the system, unsetting -&gt;preset in<br /> snd_hda_codec_cleanup_for_unbind() interferes with module unload -&gt; load<br /> scenario causing null-ptr-deref. Preset is assigned only once, during<br /> device/driver matching whereas ASoC codec driver&amp;#39;s module reloading may<br /> occur several times throughout the lifetime of an audio stack.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> btrfs: lock the inode in shared mode before starting fiemap<br /> <br /> Currently fiemap does not take the inode&amp;#39;s lock (VFS lock), it only locks<br /> a file range in the inode&amp;#39;s io tree. This however can lead to a deadlock<br /> if we have a concurrent fsync on the file and fiemap code triggers a fault<br /> when accessing the user space buffer with fiemap_fill_next_extent(). The<br /> deadlock happens on the inode&amp;#39;s i_mmap_lock semaphore, which is taken both<br /> by fsync and btrfs_page_mkwrite(). This deadlock was recently reported by<br /> syzbot and triggers a trace like the following:<br /> <br /> task:syz-executor361 state:D stack:20264 pid:5668 ppid:5119 flags:0x00004004<br /> Call Trace:<br /> <br /> context_switch kernel/sched/core.c:5293 [inline]<br /> __schedule+0x995/0xe20 kernel/sched/core.c:6606<br /> schedule+0xcb/0x190 kernel/sched/core.c:6682<br /> wait_on_state fs/btrfs/extent-io-tree.c:707 [inline]<br /> wait_extent_bit+0x577/0x6f0 fs/btrfs/extent-io-tree.c:751<br /> lock_extent+0x1c2/0x280 fs/btrfs/extent-io-tree.c:1742<br /> find_lock_delalloc_range+0x4e6/0x9c0 fs/btrfs/extent_io.c:488<br /> writepage_delalloc+0x1ef/0x540 fs/btrfs/extent_io.c:1863<br /> __extent_writepage+0x736/0x14e0 fs/btrfs/extent_io.c:2174<br /> extent_write_cache_pages+0x983/0x1220 fs/btrfs/extent_io.c:3091<br /> extent_writepages+0x219/0x540 fs/btrfs/extent_io.c:3211<br /> do_writepages+0x3c3/0x680 mm/page-writeback.c:2581<br /> filemap_fdatawrite_wbc+0x11e/0x170 mm/filemap.c:388<br /> __filemap_fdatawrite_range mm/filemap.c:421 [inline]<br /> filemap_fdatawrite_range+0x175/0x200 mm/filemap.c:439<br /> btrfs_fdatawrite_range fs/btrfs/file.c:3850 [inline]<br /> start_ordered_ops fs/btrfs/file.c:1737 [inline]<br /> btrfs_sync_file+0x4ff/0x1190 fs/btrfs/file.c:1839<br /> generic_write_sync include/linux/fs.h:2885 [inline]<br /> btrfs_do_write_iter+0xcd3/0x1280 fs/btrfs/file.c:1684<br /> call_write_iter include/linux/fs.h:2189 [inline]<br /> new_sync_write fs/read_write.c:491 [inline]<br /> vfs_write+0x7dc/0xc50 fs/read_write.c:584<br /> ksys_write+0x177/0x2a0 fs/read_write.c:637<br /> do_syscall_x64 arch/x86/entry/common.c:50 [inline]<br /> do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80<br /> entry_SYSCALL_64_after_hwframe+0x63/0xcd<br /> RIP: 0033:0x7f7d4054e9b9<br /> RSP: 002b:00007f7d404fa2f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000001<br /> RAX: ffffffffffffffda RBX: 00007f7d405d87a0 RCX: 00007f7d4054e9b9<br /> RDX: 0000000000000090 RSI: 0000000020000000 RDI: 0000000000000006<br /> RBP: 00007f7d405a51d0 R08: 0000000000000000 R09: 0000000000000000<br /> R10: 0000000000000000 R11: 0000000000000246 R12: 61635f65646f6e69<br /> R13: 65646f7475616f6e R14: 7261637369646f6e R15: 00007f7d405d87a8<br /> <br /> INFO: task syz-executor361:5697 blocked for more than 145 seconds.<br /> Not tainted 6.2.0-rc3-syzkaller-00376-g7c6984405241 #0<br /> "echo 0 &gt; /proc/sys/kernel/hung_task_timeout_secs" disables this message.<br /> task:syz-executor361 state:D stack:21216 pid:5697 ppid:5119 flags:0x00004004<br /> Call Trace:<br /> <br /> context_switch kernel/sched/core.c:5293 [inline]<br /> __schedule+0x995/0xe20 kernel/sched/core.c:6606<br /> schedule+0xcb/0x190 kernel/sched/core.c:6682<br /> rwsem_down_read_slowpath+0x5f9/0x930 kernel/locking/rwsem.c:1095<br /> __down_read_common+0x54/0x2a0 kernel/locking/rwsem.c:1260<br /> btrfs_page_mkwrite+0x417/0xc80 fs/btrfs/inode.c:8526<br /> do_page_mkwrite+0x19e/0x5e0 mm/memory.c:2947<br /> wp_page_shared+0x15e/0x380 mm/memory.c:3295<br /> handle_pte_fault mm/memory.c:4949 [inline]<br /> __handle_mm_fault mm/memory.c:5073 [inline]<br /> handle_mm_fault+0x1b79/0x26b0 mm/memory.c:5219<br /> do_user_addr_fault+0x69b/0xcb0 arch/x86/mm/fault.c:1428<br /> handle_page_fault arch/x86/mm/fault.c:1519 [inline]<br /> exc_page_fault+0x7a/0x110 arch/x86/mm/fault.c:1575<br /> asm_exc_page_fault+0x22/0x30 arch/x86/include/asm/idtentry.h:570<br /> RIP: 0010:copy_user_short_string+0xd/0x40 arch/x86/lib/copy_user_64.S:233<br /> Code: 74 0a 89 (...)<br /> RSP: 0018:ffffc9000570f330 EFLAGS: 000502<br /> ---truncated---

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> drm/amdgpu/fence: Fix oops due to non-matching drm_sched init/fini<br /> <br /> Currently amdgpu calls drm_sched_fini() from the fence driver sw fini<br /> routine - such function is expected to be called only after the<br /> respective init function - drm_sched_init() - was executed successfully.<br /> <br /> Happens that we faced a driver probe failure in the Steam Deck<br /> recently, and the function drm_sched_fini() was called even without<br /> its counter-part had been previously called, causing the following oops:<br /> <br /> amdgpu: probe of 0000:04:00.0 failed with error -110<br /> BUG: kernel NULL pointer dereference, address: 0000000000000090<br /> PGD 0 P4D 0<br /> Oops: 0002 [#1] PREEMPT SMP NOPTI<br /> CPU: 0 PID: 609 Comm: systemd-udevd Not tainted 6.2.0-rc3-gpiccoli #338<br /> Hardware name: Valve Jupiter/Jupiter, BIOS F7A0113 11/04/2022<br /> RIP: 0010:drm_sched_fini+0x84/0xa0 [gpu_sched]<br /> [...]<br /> Call Trace:<br /> <br /> amdgpu_fence_driver_sw_fini+0xc8/0xd0 [amdgpu]<br /> amdgpu_device_fini_sw+0x2b/0x3b0 [amdgpu]<br /> amdgpu_driver_release_kms+0x16/0x30 [amdgpu]<br /> devm_drm_dev_init_release+0x49/0x70<br /> [...]<br /> <br /> To prevent that, check if the drm_sched was properly initialized for a<br /> given ring before calling its fini counter-part.<br /> <br /> Notice ideally we&amp;#39;d use sched.ready for that; such field is set as the latest<br /> thing on drm_sched_init(). But amdgpu seems to "override" the meaning of such<br /> field - in the above oops for example, it was a GFX ring causing the crash, and<br /> the sched.ready field was set to true in the ring init routine, regardless of<br /> the state of the DRM scheduler. Hence, we ended-up using sched.ops as per<br /> Christian&amp;#39;s suggestion [0], and also removed the no_scheduler check [1].<br /> <br /> [0] https://lore.kernel.org/amd-gfx/984ee981-2906-0eaf-ccec-9f80975cb136@amd.com/<br /> [1] https://lore.kernel.org/amd-gfx/cd0e2994-f85f-d837-609f-7056d5fb7231@amd.com/

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> Fix page corruption caused by racy check in __free_pages<br /> <br /> When we upgraded our kernel, we started seeing some page corruption like<br /> the following consistently:<br /> <br /> BUG: Bad page state in process ganesha.nfsd pfn:1304ca<br /> page:0000000022261c55 refcount:0 mapcount:-128 mapping:0000000000000000 index:0x0 pfn:0x1304ca<br /> flags: 0x17ffffc0000000()<br /> raw: 0017ffffc0000000 ffff8a513ffd4c98 ffffeee24b35ec08 0000000000000000<br /> raw: 0000000000000000 0000000000000001 00000000ffffff7f 0000000000000000<br /> page dumped because: nonzero mapcount<br /> CPU: 0 PID: 15567 Comm: ganesha.nfsd Kdump: loaded Tainted: P B O 5.10.158-1.nutanix.20221209.el7.x86_64 #1<br /> Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 04/05/2016<br /> Call Trace:<br /> dump_stack+0x74/0x96<br /> bad_page.cold+0x63/0x94<br /> check_new_page_bad+0x6d/0x80<br /> rmqueue+0x46e/0x970<br /> get_page_from_freelist+0xcb/0x3f0<br /> ? _cond_resched+0x19/0x40<br /> __alloc_pages_nodemask+0x164/0x300<br /> alloc_pages_current+0x87/0xf0<br /> skb_page_frag_refill+0x84/0x110<br /> ...<br /> <br /> Sometimes, it would also show up as corruption in the free list pointer<br /> and cause crashes.<br /> <br /> After bisecting the issue, we found the issue started from commit<br /> e320d3012d25 ("mm/page_alloc.c: fix freeing non-compound pages"):<br /> <br /> if (put_page_testzero(page))<br /> free_the_page(page, order);<br /> else if (!PageHead(page))<br /> while (order-- &gt; 0)<br /> free_the_page(page + (1

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> powerpc/64s/interrupt: Fix interrupt exit race with security mitigation switch<br /> <br /> The RFI and STF security mitigation options can flip the<br /> interrupt_exit_not_reentrant static branch condition concurrently with<br /> the interrupt exit code which tests that branch.<br /> <br /> Interrupt exit tests this condition to set MSR[EE|RI] for exit, then<br /> again in the case a soft-masked interrupt is found pending, to recover<br /> the MSR so the interrupt can be replayed before attempting to exit<br /> again. If the condition changes between these two tests, the MSR and irq<br /> soft-mask state will become corrupted, leading to warnings and possible<br /> crashes. For example, if the branch is initially true then false,<br /> MSR[EE] will be 0 but PACA_IRQ_HARD_DIS clear and EE may not get<br /> enabled, leading to warnings in irq_64.c.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> lib/generic-radix-tree.c: Don&amp;#39;t overflow in peek()<br /> <br /> When we started spreading new inode numbers throughout most of the 64<br /> bit inode space, that triggered some corner case bugs, in particular<br /> some integer overflows related to the radix tree code. Oops.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> vdpa: ifcvf: Do proper cleanup if IFCVF init fails<br /> <br /> ifcvf_mgmt_dev leaks memory if it is not freed before<br /> returning. Call is made to correct return statement<br /> so memory does not leak. ifcvf_init_hw does not take<br /> care of this so it is needed to do it here.