On Thu, May 05, 2022 at 01:26:45PM -0500, Eric W. Biederman wrote:
> From: Peter Zijlstra <peterz@infradead.org>
> 
> Currently ptrace_stop() / do_signal_stop() rely on the special states
> TASK_TRACED and TASK_STOPPED resp. to keep unique state. That is, this
> state exists only in task->__state and nowhere else.
> 
> There's two spots of bother with this:
> 
>  - PREEMPT_RT has task->saved_state which complicates matters,
>    meaning task_is_{traced,stopped}() needs to check an additional
>    variable.
> 
>  - An alternative freezer implementation that itself relies on a
>    special TASK state would loose TASK_TRACED/TASK_STOPPED and will
>    result in misbehaviour.
> 
> As such, add additional state to task->jobctl to track this state
> outside of task->__state.
> 
> NOTE: this doesn't actually fix anything yet, just adds extra state.
> 
> --EWB
>   * didn't add a unnecessary newline in signal.h
>   * Update t->jobctl in signal_wake_up and ptrace_signal_wake_up
>     instead of in signal_wake_up_state.  This prevents the clearing
>     of TASK_STOPPED and TASK_TRACED from getting lost.
>   * Added warnings if JOBCTL_STOPPED or JOBCTL_TRACED are not cleared

Hi Eric, Peter,

On s390 this patch triggers warning at kernel/ptrace.c:272 when
kill_child testcase from strace tool is repeatedly used (the source
is attached for reference):

while :; do
	strace -f -qq -e signal=none -e trace=sched_yield,/kill ./kill_child
done

It normally takes few minutes to cause the warning in -rc3, but FWIW
it hits almost immediately for ptrace_stop-cleanup-for-v5.19 tag of
git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.

Commit 7b0fe1367ef2 ("ptrace: Document that wait_task_inactive can't
fail") suggests this WARN_ON_ONCE() is not really expected, yet we
observe a child in __TASK_TRACED state. Could you please comment here?

Thanks!