====================================================== WARNING: possible circular locking dependency detected syzkaller #0 Not tainted ------------------------------------------------------ kworker/0:4/4342 is trying to acquire lock: ffffffff8ed6b610 (rds_trans_sem){++++}-{3:3}, at: rds_trans_get_preferred+0x12f/0x3a0 net/rds/transport.c:100 but task is already holding lock: ffff888058961420 (&id_priv->handler_mutex/1){+.+.}-{3:3}, at: cma_ib_req_handler+0x25ca/0x5140 drivers/infiniband/core/cma.c:2407 which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #2 (&id_priv->handler_mutex/1){+.+.}-{3:3}: __mutex_lock_common kernel/locking/mutex.c:603 [inline] __mutex_lock+0x12d/0xae0 kernel/locking/mutex.c:747 cma_ib_req_handler+0x25ca/0x5140 drivers/infiniband/core/cma.c:2407 cm_process_work+0x4b/0x410 drivers/infiniband/core/cm.c:1854 cm_req_handler drivers/infiniband/core/cm.c:2220 [inline] cm_work_handler+0x9ca6/0xf0f0 drivers/infiniband/core/cm.c:3869 process_one_work+0x8ab/0x1160 kernel/workqueue.c:2292 worker_thread+0xaf5/0x12a0 kernel/workqueue.c:2439 kthread+0x29d/0x330 kernel/kthread.c:376 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:295 -> #1 (&id_priv->handler_mutex){+.+.}-{3:3}: __mutex_lock_common kernel/locking/mutex.c:603 [inline] __mutex_lock+0x12d/0xae0 kernel/locking/mutex.c:747 rdma_destroy_id+0x17/0x20 drivers/infiniband/core/cma.c:2080 rds_ib_laddr_check_cm net/rds/ib.c:491 [inline] rds_ib_laddr_check+0x3d0/0x6a0 net/rds/ib.c:513 rds_trans_get_preferred+0x16f/0x3a0 net/rds/transport.c:104 rds_bind+0x68c/0x870 net/rds/bind.c:254 __sys_bind+0x2e9/0x3e0 net/socket.c:1811 __do_sys_bind net/socket.c:1822 [inline] __se_sys_bind net/socket.c:1820 [inline] __x64_sys_bind+0x76/0x80 net/socket.c:1820 do_syscall_x64 arch/x86/entry/common.c:46 [inline] do_syscall_64+0x4c/0xa0 arch/x86/entry/common.c:76 entry_SYSCALL_64_after_hwframe+0x68/0xd2 -> #0 (rds_trans_sem){++++}-{3:3}: check_prev_add kernel/locking/lockdep.c:3162 [inline] check_prevs_add kernel/locking/lockdep.c:3281 [inline] validate_chain kernel/locking/lockdep.c:3897 [inline] __lock_acquire+0x3a03/0x7ea0 kernel/locking/lockdep.c:5171 lock_acquire+0x1bc/0x4a0 kernel/locking/lockdep.c:5790 down_read+0x42/0x2d0 kernel/locking/rwsem.c:1520 rds_trans_get_preferred+0x12f/0x3a0 net/rds/transport.c:100 __rds_conn_create+0xa1a/0x1fe0 net/rds/connection.c:250 rds_conn_create+0x3f/0x50 net/rds/connection.c:364 rds_ib_cm_handle_connect+0x474/0x10d0 net/rds/ib_cm.c:860 rds_rdma_cm_event_handler_cmn+0x454/0x8e0 net/rds/rdma_transport.c:85 cma_cm_event_handler+0xa6/0x310 drivers/infiniband/core/cma.c:2137 cma_ib_req_handler+0x2cfc/0x5140 drivers/infiniband/core/cma.c:2418 cm_process_work+0x4b/0x410 drivers/infiniband/core/cm.c:1854 cm_req_handler drivers/infiniband/core/cm.c:2220 [inline] cm_work_handler+0x9ca6/0xf0f0 drivers/infiniband/core/cm.c:3869 process_one_work+0x8ab/0x1160 kernel/workqueue.c:2292 worker_thread+0xaf5/0x12a0 kernel/workqueue.c:2439 kthread+0x29d/0x330 kernel/kthread.c:376 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:295 other info that might help us debug this: Chain exists of: rds_trans_sem --> &id_priv->handler_mutex --> &id_priv->handler_mutex/1 Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&id_priv->handler_mutex/1); lock(&id_priv->handler_mutex); lock(&id_priv->handler_mutex/1); rlock(rds_trans_sem); *** DEADLOCK *** 4 locks held by kworker/0:4/4342: #0: ffff88814b506938 ((wq_completion)ib_cm){+.+.}-{0:0}, at: process_one_work+0x7b4/0x1160 kernel/workqueue.c:2267 #1: ffffc900041c7d00 ((work_completion)(&(&work->work)->work)){+.+.}-{0:0}, at: process_one_work+0x7b4/0x1160 kernel/workqueue.c:2267 #2: ffff88802982f420 (&id_priv->handler_mutex){+.+.}-{3:3}, at: cma_ib_req_handler+0x20d6/0x5140 drivers/infiniband/core/cma.c:2384 #3: ffff888058961420 (&id_priv->handler_mutex/1){+.+.}-{3:3}, at: cma_ib_req_handler+0x25ca/0x5140 drivers/infiniband/core/cma.c:2407 stack backtrace: CPU: 0 PID: 4342 Comm: kworker/0:4 Not tainted syzkaller #0 Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/16/2026 Workqueue: ib_cm cm_work_handler Call Trace: dump_stack_lvl+0x188/0x24e lib/dump_stack.c:106 check_noncircular+0x2f8/0x400 kernel/locking/lockdep.c:2215 check_prev_add kernel/locking/lockdep.c:3162 [inline] check_prevs_add kernel/locking/lockdep.c:3281 [inline] validate_chain kernel/locking/lockdep.c:3897 [inline] __lock_acquire+0x3a03/0x7ea0 kernel/locking/lockdep.c:5171 lock_acquire+0x1bc/0x4a0 kernel/locking/lockdep.c:5790 down_read+0x42/0x2d0 kernel/locking/rwsem.c:1520 rds_trans_get_preferred+0x12f/0x3a0 net/rds/transport.c:100 __rds_conn_create+0xa1a/0x1fe0 net/rds/connection.c:250 rds_conn_create+0x3f/0x50 net/rds/connection.c:364 rds_ib_cm_handle_connect+0x474/0x10d0 net/rds/ib_cm.c:860 rds_rdma_cm_event_handler_cmn+0x454/0x8e0 net/rds/rdma_transport.c:85 cma_cm_event_handler+0xa6/0x310 drivers/infiniband/core/cma.c:2137 cma_ib_req_handler+0x2cfc/0x5140 drivers/infiniband/core/cma.c:2418 cm_process_work+0x4b/0x410 drivers/infiniband/core/cm.c:1854 cm_req_handler drivers/infiniband/core/cm.c:2220 [inline] cm_work_handler+0x9ca6/0xf0f0 drivers/infiniband/core/cm.c:3869 process_one_work+0x8ab/0x1160 kernel/workqueue.c:2292 worker_thread+0xaf5/0x12a0 kernel/workqueue.c:2439 kthread+0x29d/0x330 kernel/kthread.c:376 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:295 RDS/IB: Passive conn connected <::ffff:172.20.20.170,::ffff:172.20.20.170,0> version 3.1