CVE-2025-38211

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> RDMA/iwcm: Fix use-after-free of work objects after cm_id destruction<br /> <br /> The commit 59c68ac31e15 ("iw_cm: free cm_id resources on the last<br /> deref") simplified cm_id resource management by freeing cm_id once all<br /> references to the cm_id were removed. The references are removed either<br /> upon completion of iw_cm event handlers or when the application destroys<br /> the cm_id. This commit introduced the use-after-free condition where<br /> cm_id_private object could still be in use by event handler works during<br /> the destruction of cm_id. The commit aee2424246f9 ("RDMA/iwcm: Fix a<br /> use-after-free related to destroying CM IDs") addressed this use-after-<br /> free by flushing all pending works at the cm_id destruction.<br /> <br /> However, still another use-after-free possibility remained. It happens<br /> with the work objects allocated for each cm_id_priv within<br /> alloc_work_entries() during cm_id creation, and subsequently freed in<br /> dealloc_work_entries() once all references to the cm_id are removed.<br /> If the cm_id&amp;#39;s last reference is decremented in the event handler work,<br /> the work object for the work itself gets removed, and causes the use-<br /> after-free BUG below:<br /> <br /> BUG: KASAN: slab-use-after-free in __pwq_activate_work+0x1ff/0x250<br /> Read of size 8 at addr ffff88811f9cf800 by task kworker/u16:1/147091<br /> <br /> CPU: 2 UID: 0 PID: 147091 Comm: kworker/u16:1 Not tainted 6.15.0-rc2+ #27 PREEMPT(voluntary)<br /> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.3-3.fc41 04/01/2014<br /> Workqueue: 0x0 (iw_cm_wq)<br /> Call Trace:<br /> <br /> dump_stack_lvl+0x6a/0x90<br /> print_report+0x174/0x554<br /> ? __virt_addr_valid+0x208/0x430<br /> ? __pwq_activate_work+0x1ff/0x250<br /> kasan_report+0xae/0x170<br /> ? __pwq_activate_work+0x1ff/0x250<br /> __pwq_activate_work+0x1ff/0x250<br /> pwq_dec_nr_in_flight+0x8c5/0xfb0<br /> process_one_work+0xc11/0x1460<br /> ? __pfx_process_one_work+0x10/0x10<br /> ? assign_work+0x16c/0x240<br /> worker_thread+0x5ef/0xfd0<br /> ? __pfx_worker_thread+0x10/0x10<br /> kthread+0x3b0/0x770<br /> ? __pfx_kthread+0x10/0x10<br /> ? rcu_is_watching+0x11/0xb0<br /> ? _raw_spin_unlock_irq+0x24/0x50<br /> ? rcu_is_watching+0x11/0xb0<br /> ? __pfx_kthread+0x10/0x10<br /> ret_from_fork+0x30/0x70<br /> ? __pfx_kthread+0x10/0x10<br /> ret_from_fork_asm+0x1a/0x30<br /> <br /> <br /> Allocated by task 147416:<br /> kasan_save_stack+0x2c/0x50<br /> kasan_save_track+0x10/0x30<br /> __kasan_kmalloc+0xa6/0xb0<br /> alloc_work_entries+0xa9/0x260 [iw_cm]<br /> iw_cm_connect+0x23/0x4a0 [iw_cm]<br /> rdma_connect_locked+0xbfd/0x1920 [rdma_cm]<br /> nvme_rdma_cm_handler+0x8e5/0x1b60 [nvme_rdma]<br /> cma_cm_event_handler+0xae/0x320 [rdma_cm]<br /> cma_work_handler+0x106/0x1b0 [rdma_cm]<br /> process_one_work+0x84f/0x1460<br /> worker_thread+0x5ef/0xfd0<br /> kthread+0x3b0/0x770<br /> ret_from_fork+0x30/0x70<br /> ret_from_fork_asm+0x1a/0x30<br /> <br /> Freed by task 147091:<br /> kasan_save_stack+0x2c/0x50<br /> kasan_save_track+0x10/0x30<br /> kasan_save_free_info+0x37/0x60<br /> __kasan_slab_free+0x4b/0x70<br /> kfree+0x13a/0x4b0<br /> dealloc_work_entries+0x125/0x1f0 [iw_cm]<br /> iwcm_deref_id+0x6f/0xa0 [iw_cm]<br /> cm_work_handler+0x136/0x1ba0 [iw_cm]<br /> process_one_work+0x84f/0x1460<br /> worker_thread+0x5ef/0xfd0<br /> kthread+0x3b0/0x770<br /> ret_from_fork+0x30/0x70<br /> ret_from_fork_asm+0x1a/0x30<br /> <br /> Last potentially related work creation:<br /> kasan_save_stack+0x2c/0x50<br /> kasan_record_aux_stack+0xa3/0xb0<br /> __queue_work+0x2ff/0x1390<br /> queue_work_on+0x67/0xc0<br /> cm_event_handler+0x46a/0x820 [iw_cm]<br /> siw_cm_upcall+0x330/0x650 [siw]<br /> siw_cm_work_handler+0x6b9/0x2b20 [siw]<br /> process_one_work+0x84f/0x1460<br /> worker_thread+0x5ef/0xfd0<br /> kthread+0x3b0/0x770<br /> ret_from_fork+0x30/0x70<br /> ret_from_fork_asm+0x1a/0x30<br /> <br /> This BUG is reproducible by repeating the blktests test case nvme/061<br /> for the rdma transport and the siw driver.<br /> <br /> To avoid the use-after-free of cm_id_private work objects, ensure that<br /> the last reference to the cm_id is decremented not in the event handler<br /> works, but in the cm_id destruction context. For that purpose, mo<br /> ---truncated---