Performing verification for noble. I started a fresh VM and installed the latest 6.8.0-78-generic kernel from -updates. I ran the syzkaller reproducer, and after a few minutes dumped stuck processes:
$ echo w > /proc/sysrq-trigger $ dmesg kernel: INFO: task rcu_tasks_trace:15 blocked for more than 121 seconds. kernel: Not tainted 6.8.0-78-generic #78-Ubuntu task:rcu_tasks_trace state:D stack:0 pid:15 tgid:15 ppid:2 flags:0x00004000 Call Trace: <TASK> __schedule+0x27c/0x6b0 schedule+0x33/0x110 schedule_timeout+0x157/0x170 wait_for_completion+0x88/0x150 __wait_rcu_gp+0x17e/0x190 synchronize_rcu+0x12d/0x140 ? __pfx_call_rcu_hurry+0x10/0x10 ? __pfx_wakeme_after_rcu+0x10/0x10 rcu_tasks_trace_postscan+0xe/0x20 rcu_tasks_wait_gp+0x119/0x310 ? _raw_spin_lock_irqsave+0xe/0x20 ? rcu_tasks_need_gpcb+0x1f7/0x350 ? __pfx_rcu_tasks_kthread+0x10/0x10 rcu_tasks_one_gp+0x122/0x150 rcu_tasks_kthread+0xa4/0xd0 kthread+0xef/0x120 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x44/0x70 ? __pfx_kthread+0x10/0x10 ret_from_fork_asm+0x1b/0x30 </TASK> task:system-probe state:D stack:0 pid:7105 tgid:7021 ppid:6579 flags:0x00004006 Call Trace: <TASK> __schedule+0x27c/0x6b0 schedule+0x33/0x110 schedule_timeout+0x157/0x170 wait_for_completion+0x88/0x150 __wait_rcu_gp+0x17e/0x190 synchronize_rcu_tasks_generic+0x64/0xe0 ? __pfx_call_rcu_tasks_trace+0x10/0x10 ? __pfx_wakeme_after_rcu+0x10/0x10 synchronize_rcu_tasks_trace+0x15/0x20 perf_event_detach_bpf_prog+0x7d/0xe0 _free_event+0x20e/0x2a0 perf_event_release_kernel+0x281/0x2e0 ? __fput+0x160/0x2e0 perf_release+0x15/0x30 __fput+0xa0/0x2e0 ____fput+0xe/0x20 task_work_run+0x5e/0xa0 do_exit+0x2b6/0x530 do_group_exit+0x35/0x90 get_signal+0x96e/0x9b0 arch_do_signal_or_restart+0x39/0x120 syscall_exit_to_user_mode+0x206/0x260 do_syscall_64+0x8c/0x180 ? handle_mm_fault+0xad/0x380 ? do_user_addr_fault+0x334/0x670 ? irqentry_exit_to_user_mode+0x7b/0x260 ? irqentry_exit+0x43/0x50 ? exc_page_fault+0x94/0x1b0 entry_SYSCALL_64_after_hwframe+0x78/0x80 I then enabled -proposed, and installed 6.8.0-80-generic and rebooted. From there, I ran the syzkaller reproducer for a very long time. Things kept running, and there were no hung task timeouts. The kernel in -proposed appears to fix the problem, as it does not reproduce in a very long time. Happy to mark verified. ** Tags removed: verification-needed-noble-linux ** Tags added: verification-done-noble-linux -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2117123 Title: rcu: Eliminate deadlocks involving do_exit() and RCU tasks To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2117123/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
