https://bugzilla.kernel.org/show_bug.cgi?id=214147 --- Comment #4 from Badalian Slava (slavon.net@xxxxxxxxx) --- I get some info . Test case have 2 zfs pool. HDD is zdev (raid6) SDD is stype. simple copy from SSD to HDD Specific^ 1. I use NFS with RDMA and 16 threads (for linux hosts) 2/ i use targetcli and nvmet for share devs to WinClusteredHosts When se get deadlock i see in perf amples: 1M of event 'cycles', Event count (approx.): 711199023386 Children Self Command Shared Object Symbol - 30,57% 0,00% z_wr_iss [kernel.kallsyms] [k] ret_from_fork ◆ ret_from_fork ▒ - kthread ▒ - 30,53% taskq_create ▒ - 16,50% zio_inject_init ▒ - 4,77% 0xffffffffc0c7285f ▒ - 4,73% raidz_rec_pr_coeff.isra.5 ▒ - 3,86% secpolicy_vnode_setid_retain ▒ - 2,10% zthr_resume ▒ - 1,85% vdev_label_init ▒ - 1,03% secpolicy_vnode_setid_retain ▒ - 0,73% zthr_cancel ▒ 0,72% vdev_rebuild_update_sync ▒ - 1,57% zthr_cancel ▒ - 1,55% vdev_rebuild_update_sync ▒ 0,53% __mutex_lock.isra.13 ▒ 0,53% avl_remove ▒ 0,71% abd_bio_map_off ▒ - 3,70% vdev_disk_open ▒ - 3,69% range_tree_remove_impl ▒ - 3,45% range_tree_add_impl ▒ - 2,66% mmp_thread ▒ 1,24% metaslab_enable ▒ 0,74% __mutex_lock.isra.13 ▒ - 2,71% zvol_set_volsize ▒ - 2,51% zfs_aclset_common ▒ - 2,51% metaslab_load.part.29 ▒ 1,31% metaslab_class_evict_old ▒ 0,96% metaslab_unload.part.28 ▒ - 1,84% zvol_set_volmode_sync_cb ▒ - 1,83% zio_handle_device_injection_impl.part.15 ▒ - 1,57% spl_taskq_fini ▒ - 1,55% _raw_spin_lock_irqsave ▒ native_queued_spin_lock_slowpath ▒ - 1,20% zvol_set_volmode_sync ▒ 1,02% l2arc_read_done ▒ - 0,63% zio_inject_list_next ▒ - 0,62% zfs_mode_compute ▒ - 0,59% zfs_mode_compute ▒ 0,53% abd_iterate_func2.part.16 ▒ - 12,00% _raw_spin_lock_irqsave ▒ native_queued_spin_lock_slowpath ▒ - 0,93% schedule ▒ 0,86% __sched_text_start ▒ - 30,57% 0,00% z_wr_iss [kernel.kallsyms] [k] kthread ▒ - 30,57% kthread ▒ - 30,53% taskq_create ▒ - 16,50% zio_inject_init ▒ - 4,77% 0xffffffffc0c7285f ▒ - 4,73% raidz_rec_pr_coeff.isra.5 and i see in [root@vm2 ~]# cat /proc/locks 1: FLOCK ADVISORY WRITE 999321 00:18:3499 0 EOF 2: POSIX ADVISORY WRITE 949090 00:18:3395 0 EOF 3: POSIX ADVISORY WRITE 949090 00:18:3394 0 EOF 4: POSIX ADVISORY WRITE 948911 00:18:3322 0 EOF 5: POSIX ADVISORY WRITE 999305 00:18:3511 0 EOF butt one.one/two in ACTIVE state., pid is nfs5 kernel p[process. deadlocked in nfs! -- You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
