Vulnerabilities

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> Revert "sched/fair: Make sure to try to detach at least one movable task"<br /> <br /> This reverts commit b0defa7ae03ecf91b8bfd10ede430cff12fcbd06.<br /> <br /> b0defa7ae03ec changed the load balancing logic to ignore env.max_loop if<br /> all tasks examined to that point were pinned. The goal of the patch was<br /> to make it more likely to be able to detach a task buried in a long list<br /> of pinned tasks. However, this has the unfortunate side effect of<br /> creating an O(n) iteration in detach_tasks(), as we now must fully<br /> iterate every task on a cpu if all or most are pinned. Since this load<br /> balance code is done with rq lock held, and often in softirq context, it<br /> is very easy to trigger hard lockups. We observed such hard lockups with<br /> a user who affined O(10k) threads to a single cpu.<br /> <br /> When I discussed this with Vincent he initially suggested that we keep<br /> the limit on the number of tasks to detach, but increase the number of<br /> tasks we can search. However, after some back and forth on the mailing<br /> list, he recommended we instead revert the original patch, as it seems<br /> likely no one was actually getting hit by the original issue.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> net, sunrpc: Remap EPERM in case of connection failure in xs_tcp_setup_socket<br /> <br /> When using a BPF program on kernel_connect(), the call can return -EPERM. This<br /> causes xs_tcp_setup_socket() to loop forever, filling up the syslog and causing<br /> the kernel to potentially freeze up.<br /> <br /> Neil suggested:<br /> <br /> This will propagate -EPERM up into other layers which might not be ready<br /> to handle it. It might be safer to map EPERM to an error we would be more<br /> likely to expect from the network system - such as ECONNREFUSED or ENETDOWN.<br /> <br /> ECONNREFUSED as error seems reasonable. For programs setting a different error<br /> can be out of reach (see handling in 4fbac77d2d09) in particular on kernels<br /> which do not have f10d05966196 ("bpf: Make BPF_PROG_RUN_ARRAY return -err<br /> instead of allow boolean"), thus given that it is better to simply remap for<br /> consistent behavior. UDP does handle EPERM in xs_udp_send_request().

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> wireguard: allowedips: avoid unaligned 64-bit memory accesses<br /> <br /> On the parisc platform, the kernel issues kernel warnings because<br /> swap_endian() tries to load a 128-bit IPv6 address from an unaligned<br /> memory location:<br /> <br /> Kernel: unaligned access to 0x55f4688c in wg_allowedips_insert_v6+0x2c/0x80 [wireguard] (iir 0xf3010df)<br /> Kernel: unaligned access to 0x55f46884 in wg_allowedips_insert_v6+0x38/0x80 [wireguard] (iir 0xf2010dc)<br /> <br /> Avoid such unaligned memory accesses by instead using the<br /> get_unaligned_be64() helper macro.<br /> <br /> [Jason: replace src[8] in original patch with src+8]

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> cachefiles: add missing lock protection when polling<br /> <br /> Add missing lock protection in poll routine when iterating xarray,<br /> otherwise:<br /> <br /> Even with RCU read lock held, only the slot of the radix tree is<br /> ensured to be pinned there, while the data structure (e.g. struct<br /> cachefiles_req) stored in the slot has no such guarantee. The poll<br /> routine will iterate the radix tree and dereference cachefiles_req<br /> accordingly. Thus RCU read lock is not adequate in this case and<br /> spinlock is needed here.

An issue in the Hardware info module of IT Solutions Enjay CRM OS v1.0 allows attackers to escape the restricted terminal environment and gain root-level privileges on the underlying system.

An IP Spoofing vulnerability has been discovered in Likeshop up to 2.5.7.20210811. This issue allows an attacker to replace their real IP address with any arbitrary IP address, specifically by adding a forged &amp;#39;X-Forwarded&amp;#39; or &amp;#39;Client-IP&amp;#39; header to requests. Exploiting IP spoofing, attackers can bypass account lockout mechanisms during attempts to log into admin accounts, spoof IP addresses in requests sent to the server, and impersonate IP addresses that have logged into user accounts, etc.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> filemap: replace pte_offset_map() with pte_offset_map_nolock()<br /> <br /> The vmf-&gt;ptl in filemap_fault_recheck_pte_none() is still set from<br /> handle_pte_fault(). But at the same time, we did a pte_unmap(vmf-&gt;pte). <br /> After a pte_unmap(vmf-&gt;pte) unmap and rcu_read_unlock(), the page table<br /> may be racily changed and vmf-&gt;ptl maybe fails to protect the actual page<br /> table. Fix this by replacing pte_offset_map() with<br /> pte_offset_map_nolock().<br /> <br /> As David said, the PTL pointer might be stale so if we continue to use<br /> it infilemap_fault_recheck_pte_none(), it might trigger UAF. Also, if<br /> the PTL fails, the issue fixed by commit 58f327f2ce80 ("filemap: avoid<br /> unnecessary major faults in filemap_fault()") might reappear.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> mm: fix crashes from deferred split racing folio migration<br /> <br /> Even on 6.10-rc6, I&amp;#39;ve been seeing elusive "Bad page state"s (often on<br /> flags when freeing, yet the flags shown are not bad: PG_locked had been<br /> set and cleared??), and VM_BUG_ON_PAGE(page_ref_count(page) == 0)s from<br /> deferred_split_scan()&amp;#39;s folio_put(), and a variety of other BUG and WARN<br /> symptoms implying double free by deferred split and large folio migration.<br /> <br /> 6.7 commit 9bcef5973e31 ("mm: memcg: fix split queue list crash when large<br /> folio migration") was right to fix the memcg-dependent locking broken in<br /> 85ce2c517ade ("memcontrol: only transfer the memcg data for migration"),<br /> but missed a subtlety of deferred_split_scan(): it moves folios to its own<br /> local list to work on them without split_queue_lock, during which time<br /> folio-&gt;_deferred_list is not empty, but even the "right" lock does nothing<br /> to secure the folio and the list it is on.<br /> <br /> Fortunately, deferred_split_scan() is careful to use folio_try_get(): so<br /> folio_migrate_mapping() can avoid the race by folio_undo_large_rmappable()<br /> while the old folio&amp;#39;s reference count is temporarily frozen to 0 - adding<br /> such a freeze in the !mapping case too (originally, folio lock and<br /> unmapping and no swap cache left an anon folio unreachable, so no freezing<br /> was needed there: but the deferred split queue offers a way to reach it).

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> s390/mm: Add NULL pointer check to crst_table_free() base_crst_free()<br /> <br /> crst_table_free() used to work with NULL pointers before the conversion<br /> to ptdescs. Since crst_table_free() can be called with a NULL pointer<br /> (error handling in crst_table_upgrade() add an explicit check.<br /> <br /> Also add the same check to base_crst_free() for consistency reasons.<br /> <br /> In real life this should not happen, since order two GFP_KERNEL<br /> allocations will not fail, unless FAIL_PAGE_ALLOC is enabled and used.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> bpf: Fail bpf_timer_cancel when callback is being cancelled<br /> <br /> Given a schedule:<br /> <br /> timer1 cb timer2 cb<br /> <br /> bpf_timer_cancel(timer2); bpf_timer_cancel(timer1);<br /> <br /> Both bpf_timer_cancel calls would wait for the other callback to finish<br /> executing, introducing a lockup.<br /> <br /> Add an atomic_t count named &amp;#39;cancelling&amp;#39; in bpf_hrtimer. This keeps<br /> track of all in-flight cancellation requests for a given BPF timer.<br /> Whenever cancelling a BPF timer, we must check if we have outstanding<br /> cancellation requests, and if so, we must fail the operation with an<br /> error (-EDEADLK) since cancellation is synchronous and waits for the<br /> callback to finish executing. This implies that we can enter a deadlock<br /> situation involving two or more timer callbacks executing in parallel<br /> and attempting to cancel one another.<br /> <br /> Note that we avoid incrementing the cancelling counter for the target<br /> timer (the one being cancelled) if bpf_timer_cancel is not invoked from<br /> a callback, to avoid spurious errors. The whole point of detecting<br /> cur-&gt;cancelling and returning -EDEADLK is to not enter a busy wait loop<br /> (which may or may not lead to a lockup). This does not apply in case the<br /> caller is in a non-callback context, the other side can continue to<br /> cancel as it sees fit without running into errors.<br /> <br /> Background on prior attempts:<br /> <br /> Earlier versions of this patch used a bool &amp;#39;cancelling&amp;#39; bit and used the<br /> following pattern under timer-&gt;lock to publish cancellation status.<br /> <br /> lock(t-&gt;lock);<br /> t-&gt;cancelling = true;<br /> mb();<br /> if (cur-&gt;cancelling)<br /> return -EDEADLK;<br /> unlock(t-&gt;lock);<br /> hrtimer_cancel(t-&gt;timer);<br /> t-&gt;cancelling = false;<br /> <br /> The store outside the critical section could overwrite a parallel<br /> requests t-&gt;cancelling assignment to true, to ensure the parallely<br /> executing callback observes its cancellation status.<br /> <br /> It would be necessary to clear this cancelling bit once hrtimer_cancel<br /> is done, but lack of serialization introduced races. Another option was<br /> explored where bpf_timer_start would clear the bit when (re)starting the<br /> timer under timer-&gt;lock. This would ensure serialized access to the<br /> cancelling bit, but may allow it to be cleared before in-flight<br /> hrtimer_cancel has finished executing, such that lockups can occur<br /> again.<br /> <br /> Thus, we choose an atomic counter to keep track of all outstanding<br /> cancellation requests and use it to prevent lockups in case callbacks<br /> attempt to cancel each other while executing in parallel.

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> mm/shmem: disable PMD-sized page cache if needed<br /> <br /> For shmem files, it&amp;#39;s possible that PMD-sized page cache can&amp;#39;t be<br /> supported by xarray. For example, 512MB page cache on ARM64 when the base<br /> page size is 64KB can&amp;#39;t be supported by xarray. It leads to errors as the<br /> following messages indicate when this sort of xarray entry is split.<br /> <br /> WARNING: CPU: 34 PID: 7578 at lib/xarray.c:1025 xas_split_alloc+0xf8/0x128<br /> Modules linked in: binfmt_misc nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 \<br /> nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject \<br /> nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 \<br /> ip_set rfkill nf_tables nfnetlink vfat fat virtio_balloon drm fuse xfs \<br /> libcrc32c crct10dif_ce ghash_ce sha2_ce sha256_arm64 sha1_ce virtio_net \<br /> net_failover virtio_console virtio_blk failover dimlib virtio_mmio<br /> CPU: 34 PID: 7578 Comm: test Kdump: loaded Tainted: G W 6.10.0-rc5-gavin+ #9<br /> Hardware name: QEMU KVM Virtual Machine, BIOS edk2-20240524-1.el9 05/24/2024<br /> pstate: 83400005 (Nzcv daif +PAN -UAO +TCO +DIT -SSBS BTYPE=--)<br /> pc : xas_split_alloc+0xf8/0x128<br /> lr : split_huge_page_to_list_to_order+0x1c4/0x720<br /> sp : ffff8000882af5f0<br /> x29: ffff8000882af5f0 x28: ffff8000882af650 x27: ffff8000882af768<br /> x26: 0000000000000cc0 x25: 000000000000000d x24: ffff00010625b858<br /> x23: ffff8000882af650 x22: ffffffdfc0900000 x21: 0000000000000000<br /> x20: 0000000000000000 x19: ffffffdfc0900000 x18: 0000000000000000<br /> x17: 0000000000000000 x16: 0000018000000000 x15: 52f8004000000000<br /> x14: 0000e00000000000 x13: 0000000000002000 x12: 0000000000000020<br /> x11: 52f8000000000000 x10: 52f8e1c0ffff6000 x9 : ffffbeb9619a681c<br /> x8 : 0000000000000003 x7 : 0000000000000000 x6 : ffff00010b02ddb0<br /> x5 : ffffbeb96395e378 x4 : 0000000000000000 x3 : 0000000000000cc0<br /> x2 : 000000000000000d x1 : 000000000000000c x0 : 0000000000000000<br /> Call trace:<br /> xas_split_alloc+0xf8/0x128<br /> split_huge_page_to_list_to_order+0x1c4/0x720<br /> truncate_inode_partial_folio+0xdc/0x160<br /> shmem_undo_range+0x2bc/0x6a8<br /> shmem_fallocate+0x134/0x430<br /> vfs_fallocate+0x124/0x2e8<br /> ksys_fallocate+0x4c/0xa0<br /> __arm64_sys_fallocate+0x24/0x38<br /> invoke_syscall.constprop.0+0x7c/0xd8<br /> do_el0_svc+0xb4/0xd0<br /> el0_svc+0x44/0x1d8<br /> el0t_64_sync_handler+0x134/0x150<br /> el0t_64_sync+0x17c/0x180<br /> <br /> Fix it by disabling PMD-sized page cache when HPAGE_PMD_ORDER is larger<br /> than MAX_PAGECACHE_ORDER. As Matthew Wilcox pointed, the page cache in a<br /> shmem file isn&amp;#39;t represented by a multi-index entry and doesn&amp;#39;t have this<br /> limitation when the xarry entry is split until commit 6b24ca4a1a8d ("mm:<br /> Use multi-index entries in the page cache").

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> libceph: fix race between delayed_work() and ceph_monc_stop()<br /> <br /> The way the delayed work is handled in ceph_monc_stop() is prone to<br /> races with mon_fault() and possibly also finish_hunting(). Both of<br /> these can requeue the delayed work which wouldn&amp;#39;t be canceled by any of<br /> the following code in case that happens after cancel_delayed_work_sync()<br /> runs -- __close_session() doesn&amp;#39;t mess with the delayed work in order<br /> to avoid interfering with the hunting interval logic. This part was<br /> missed in commit b5d91704f53e ("libceph: behave in mon_fault() if<br /> cur_mon auth and monc-&gt;monmap being<br /> particularly susceptible to quickly being reused.<br /> <br /> To fix this:<br /> <br /> - clear monc-&gt;cur_mon and monc-&gt;hunting as part of closing the session<br /> in ceph_monc_stop()<br /> - bail from delayed_work() if monc-&gt;cur_mon is cleared, similar to how<br /> it&amp;#39;s done in mon_fault() and finish_hunting() (based on monc-&gt;hunting)<br /> - call cancel_delayed_work_sync() after the session is closed