CVE-2025-21983

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> mm/slab/kvfree_rcu: Switch to WQ_MEM_RECLAIM wq<br /> <br /> Currently kvfree_rcu() APIs use a system workqueue which is<br /> "system_unbound_wq" to driver RCU machinery to reclaim a memory.<br /> <br /> Recently, it has been noted that the following kernel warning can<br /> be observed:<br /> <br /> <br /> workqueue: WQ_MEM_RECLAIM nvme-wq:nvme_scan_work is flushing !WQ_MEM_RECLAIM events_unbound:kfree_rcu_work<br /> WARNING: CPU: 21 PID: 330 at kernel/workqueue.c:3719 check_flush_dependency+0x112/0x120<br /> Modules linked in: intel_uncore_frequency(E) intel_uncore_frequency_common(E) skx_edac(E) ...<br /> CPU: 21 UID: 0 PID: 330 Comm: kworker/u144:6 Tainted: G E 6.13.2-0_g925d379822da #1<br /> Hardware name: Wiwynn Twin Lakes MP/Twin Lakes Passive MP, BIOS YMM20 02/01/2023<br /> Workqueue: nvme-wq nvme_scan_work<br /> RIP: 0010:check_flush_dependency+0x112/0x120<br /> Code: 05 9a 40 14 02 01 48 81 c6 c0 00 00 00 48 8b 50 18 48 81 c7 c0 00 00 00 48 89 f9 48 ...<br /> RSP: 0018:ffffc90000df7bd8 EFLAGS: 00010082<br /> RAX: 000000000000006a RBX: ffffffff81622390 RCX: 0000000000000027<br /> RDX: 00000000fffeffff RSI: 000000000057ffa8 RDI: ffff88907f960c88<br /> RBP: 0000000000000000 R08: ffffffff83068e50 R09: 000000000002fffd<br /> R10: 0000000000000004 R11: 0000000000000000 R12: ffff8881001a4400<br /> R13: 0000000000000000 R14: ffff88907f420fb8 R15: 0000000000000000<br /> FS: 0000000000000000(0000) GS:ffff88907f940000(0000) knlGS:0000000000000000<br /> CR2: 00007f60c3001000 CR3: 000000107d010005 CR4: 00000000007726f0<br /> PKRU: 55555554<br /> Call Trace:<br /> <br /> ? __warn+0xa4/0x140<br /> ? check_flush_dependency+0x112/0x120<br /> ? report_bug+0xe1/0x140<br /> ? check_flush_dependency+0x112/0x120<br /> ? handle_bug+0x5e/0x90<br /> ? exc_invalid_op+0x16/0x40<br /> ? asm_exc_invalid_op+0x16/0x20<br /> ? timer_recalc_next_expiry+0x190/0x190<br /> ? check_flush_dependency+0x112/0x120<br /> ? check_flush_dependency+0x112/0x120<br /> __flush_work.llvm.1643880146586177030+0x174/0x2c0<br /> flush_rcu_work+0x28/0x30<br /> kvfree_rcu_barrier+0x12f/0x160<br /> kmem_cache_destroy+0x18/0x120<br /> bioset_exit+0x10c/0x150<br /> disk_release.llvm.6740012984264378178+0x61/0xd0<br /> device_release+0x4f/0x90<br /> kobject_put+0x95/0x180<br /> nvme_put_ns+0x23/0xc0<br /> nvme_remove_invalid_namespaces+0xb3/0xd0<br /> nvme_scan_work+0x342/0x490<br /> process_scheduled_works+0x1a2/0x370<br /> worker_thread+0x2ff/0x390<br /> ? pwq_release_workfn+0x1e0/0x1e0<br /> kthread+0xb1/0xe0<br /> ? __kthread_parkme+0x70/0x70<br /> ret_from_fork+0x30/0x40<br /> ? __kthread_parkme+0x70/0x70<br /> ret_from_fork_asm+0x11/0x20<br /> <br /> ---[ end trace 0000000000000000 ]---<br /> <br /> <br /> To address this switch to use of independent WQ_MEM_RECLAIM<br /> workqueue, so the rules are not violated from workqueue framework<br /> point of view.<br /> <br /> Apart of that, since kvfree_rcu() does reclaim memory it is worth<br /> to go with WQ_MEM_RECLAIM type of wq because it is designed for<br /> this purpose.