{"id":"CVE-2026-45973","title":"RDMA/mlx5: Fix UMR hang in LAG error state unload","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nRDMA/mlx5: Fix UMR hang in LAG error state unload\n\nDuring firmware reset in LAG mode, a race condition causes the driver\nto hang indefinitely while waiting for UMR comp…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= 921fcf2971a1e8d3b904ba2c2905b96f4ec3d4ad < 33deba9bf5c34335182342588315f3db7dbc56d3","Linux >= 6b0acf6a94c31efa43fce4edc22413a3390f9c05 < c8fb5c965ac7d0104872a8e4f6451f3bc6328199","Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < 6d838873da9cb97551d42316967cc82bf8f8031b","Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < 613f5d4139b6ba801ccd93f9a28943be60d903bc","Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < ebc2164a4cd4314503f1a0c8e7aaf76d7e5fa211","Linux 542bd62b7a7f37182c9ef192c2bd25d118c144e4","Linux >= 6.6.64 < 6.6.160","Linux >= 6.12.2 < 6.12.75","Linux >= 6.11.11 < 6.12","Linux 6.13"],"published":"2026-05-27","updated":"2026-10-21","sourceUpdated":"2026-10-21T12:41:08.534Z","source":"CVEORG","sourceUrl":"https://www.cve.org/CVERecord?id=CVE-2026-45973","references":[{"url":"https://git.kernel.org/stable/c/33deba9bf5c34335182342588315f3db7dbc56d3"},{"url":"https://git.kernel.org/stable/c/c8fb5c965ac7d0104872a8e4f6451f3bc6328199"},{"url":"https://git.kernel.org/stable/c/6d838873da9cb97551d42316967cc82bf8f8031b"},{"url":"https://git.kernel.org/stable/c/613f5d4139b6ba801ccd93f9a28943be60d903bc"},{"url":"https://git.kernel.org/stable/c/ebc2164a4cd4314503f1a0c8e7aaf76d7e5fa211"}],"tags":["cve.org"],"epss":0.00162,"epssPercentile":0.04979,"ingestedAt":"2026-10-21T12:48:53.632Z","slug":"CVE-2026-45973","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nRDMA/mlx5: Fix UMR hang in LAG error state unload\n\nDuring firmware reset in LAG mode, a race condition causes the driver\nto hang indefinitely while waiting for UMR completion during device\nunload. See [1].\n\nIn LAG mode the bond device is only registered on the master, so it\nnever sees sys_error events from the slave.\nDuring firmware reset this causes UMR waits to hang forever on unload\nas the slave is dead but the master hasn't entered error state yet, so\nUMR posts succeed but completions never arrive.\n\nFix this by adding a sys_error notifier that gets registered before\nMLX5_IB_STAGE_IB_REG and stays alive until after ib_unregister_device().\nThis ensures error events reach the bond device throughout teardown.\n\n[1]\nCall Trace:\n __schedule+0x2bd/0x760\n schedule+0x37/0xa0\n schedule_preempt_disabled+0xa/0x10\n __mutex_lock.isra.6+0x2b5/0x4a0\n __mlx5_ib_dereg_mr+0x606/0x870 [mlx5_ib]\n ? __xa_erase+0x4a/0xa0\n ? _cond_resched+0x15/0x30\n ? wait_for_completion+0x31/0x100\n ib_dereg_mr_user+0x48/0xc0 [ib_core]\n ? rdmacg_uncharge_hierarchy+0xa0/0x100\n destroy_hw_idr_uobject+0x20/0x50 [ib_uverbs]\n uverbs_destroy_uobject+0x37/0x150 [ib_uverbs]\n __uverbs_cleanup_ufile+0xda/0x140 [ib_uverbs]\n uverbs_destroy_ufile_hw+0x3a/0xf0 [ib_uverbs]\n ib_uverbs_remove_one+0xc3/0x140 [ib_uverbs]\n remove_client_context+0x8b/0xd0 [ib_core]\n disable_device+0x8c/0x130 [ib_core]\n __ib_unregister_device+0x10d/0x180 [ib_core]\n ib_unregister_device+0x21/0x30 [ib_core]\n __mlx5_ib_remove+0x1e4/0x1f0 [mlx5_ib]\n auxiliary_bus_remove+0x1e/0x30\n device_release_driver_internal+0x103/0x1f0\n bus_remove_device+0xf7/0x170\n device_del+0x181/0x410\n mlx5_rescan_drivers_locked.part.10+0xa9/0x1d0 [mlx5_core]\n mlx5_disable_lag+0x253/0x260 [mlx5_core]\n mlx5_lag_disable_change+0x89/0xc0 [mlx5_core]\n mlx5_eswitch_disable+0x67/0xa0 [mlx5_core]\n mlx5_unload+0x15/0xd0 [mlx5_core]\n mlx5_unload_one+0x71/0xc0 [mlx5_core]\n mlx5_sync_reset_reload_work+0x83/0x100 [mlx5_core]\n process_one_work+0x1a7/0x360\n worker_thread+0x30/0x390\n ? create_worker+0x1a0/0x1a0\n kthread+0x116/0x130\n ? kthread_flush_work_fn+0x10/0x10\n ret_from_fork+0x22/0x40\n\n## Affected\n\n- `Linux >= 921fcf2971a1e8d3b904ba2c2905b96f4ec3d4ad < 33deba9bf5c34335182342588315f3db7dbc56d3`\n- `Linux >= 6b0acf6a94c31efa43fce4edc22413a3390f9c05 < c8fb5c965ac7d0104872a8e4f6451f3bc6328199`\n- `Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < 6d838873da9cb97551d42316967cc82bf8f8031b`\n- `Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < 613f5d4139b6ba801ccd93f9a28943be60d903bc`\n- `Linux >= ede132a5cf559f3ab35a4c28bac4f4a6c20334d8 < ebc2164a4cd4314503f1a0c8e7aaf76d7e5fa211`\n- `Linux 542bd62b7a7f37182c9ef192c2bd25d118c144e4`\n- `Linux >= 6.6.64 < 6.6.160`\n- `Linux >= 6.12.2 < 6.12.75`\n- `Linux >= 6.11.11 < 6.12`\n- `Linux 6.13`\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":3,"depthScoreParts":{"impact":2.8,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}