On Wed, Jul 06, 2022 at 11:27:05AM +0200, Sven Schnelle wrote: > Peter Zijlstra <peterz@xxxxxxxxxxxxx> writes: > > > On Wed, Jul 06, 2022 at 09:58:55AM +0200, Sven Schnelle wrote: > > > >> >> [ 86.218551] kill_chi-343805 6d.... 79990141us : ptrace_stop: JOBCTL_TRACED already set, state=0 <------ valid combination of flags? > >> > > >> > Yeah, that's not supposed to be so. JOBCTL_TRACED is supposed to follow > >> > __TASK_TRACED for now. Set when __TASK_TRACED, cleared when > >> > TASK_RUNNING. > >> > > >> > Specifically {ptrace_,}signal_wake_up() in signal.h clear JOBCTL_TRACED > >> > when they would wake a __TASK_TRACED task. > >> > >> try_to_wake_up() clears TASK_TRACED in this case because a signal > >> (SIGKILL) has to be delivered. As a test I put the following change > >> on top, and it "fixes" the problem: > >> > >> diff --git a/kernel/sched/core.c b/kernel/sched/core.c > >> index da0bf6fe9ecd..f2e0f5e70e77 100644 > >> --- a/kernel/sched/core.c > >> +++ b/kernel/sched/core.c > >> @@ -4141,6 +4149,9 @@ try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) > >> * TASK_WAKING such that we can unlock p->pi_lock before doing the > >> * enqueue, such as ttwu_queue_wakelist(). > >> */ > >> + if (p->__state & TASK_TRACED) > >> + trace_printk("clearing TASK_TRACED 2\n"); > >> + p->jobctl &= ~JOBCTL_TRACED; > >> WRITE_ONCE(p->__state, TASK_WAKING); > >> > >> /* > >> > >> There are several places where the state is changed from TASK_TRACED to > >> something else without clearing JOBCTL_TRACED. > > > > I'm having difficulty spotting them; I find: > > > > TASK_WAKEKILL: signal_wake_up() > > __TASK_TRACED: ptrace_signal_wake_up(), ptrace_unfreeze_traced(), ptrace_resume() > > > > And all those sites dutifully clear JOBCTL_TRACED. > > > > I'd be most interested in the calstack for the 'clearing TASK_TRACED 2' > > events to see where we miss a spot. > > The calltrace is: > [ 9.863613] Call Trace: > [ 9.863616] [<00000000d3105f0e>] try_to_wake_up+0xae/0x620 > [ 9.863620] ([<00000000d3106164>] try_to_wake_up+0x304/0x620) > [ 9.863623] [<00000000d30d1e46>] ptrace_unfreeze_traced+0x9e/0xa8 > [ 9.863629] [<00000000d30d2ef0>] __s390x_sys_ptrace+0xc0/0x160 > [ 9.863633] [<00000000d3c5d8f4>] __do_syscall+0x1d4/0x200 > [ 9.863678] [<00000000d3c6c332>] system_call+0x82/0xb0 > [ 9.863685] Last Breaking-Event-Address: > [ 9.863686] [<00000000d3106176>] try_to_wake_up+0x316/0x620 > [ 9.863688] ---[ end trace 0000000000000000 ]--- > > ptrace_unfreeze_traced() is: > > static void ptrace_unfreeze_traced(struct task_struct *task) > { > unsigned long flags; > > /* > * The child may be awake and may have cleared > * JOBCTL_PTRACE_FROZEN (see ptrace_resume). The child will > * not set JOBCTL_PTRACE_FROZEN or enter __TASK_TRACED anew. > */ > if (lock_task_sighand(task, &flags)) { > task->jobctl &= ~JOBCTL_PTRACE_FROZEN; > if (__fatal_signal_pending(task)) { > task->jobctl &= ~TASK_TRACED; > > Looking at this, shouldn't the line above read task->jobctl &= ~JOBCTL_TRACED? YES! Absolutely. > wake_up_state(task, __TASK_TRACED); > } > unlock_task_sighand(task, &flags); > } > }