Re: [PATCH 9/9] ptrace: Don't change __state

[Oleg Nesterov]

On 04/28, Eric W. Biederman wrote:
>
> Oleg Nesterov <oleg@xxxxxxxxxx> writes:
>
> >> The bug appears when the TRACEE makes it to schedule().  Inside
> >> schedule there is a call to signal_pending_state() which notices
> >> a SIGKILL is pending and refuses to sleep.
> >
> > And I think this is fine. This doesn't really differ from the case
> > when the tracee was killed before it takes siglock.
>
> Hmm.  Maybe.

I hope ;)

> Previously we were actually guaranteed in ptrace_check_attach that after
> ptrace_freeze_traced would succeed as any pending fatal signal would
> cause ptrace_freeze_traced to fail.  Any incoming fatal signal would not
> stop schedule from sleeping.

Yes.

So let me repeat, 7/9 "ptrace: Simplify the wait_task_inactive call in
ptrace_check_attach" looks good to me (except it should use
wait_task_inactive(__TASK_TRACED)), but it should come before other
meaningfull changes and the changelog should be updated.

And then we will probably need to reconsider this wait_task_inactive()
and WARN_ON() around it, but depends on what will we finally do.

> I think in my tired mind yesterday

I got lost too ;)

> Still I would like to be able to
> let wait_task_inactive not care about the state of the process it is
> going to sleep for.

Not sure... but to be honest I didn't really pay attention to the
wait_task_inactive(match_state => 0) part...

Oleg.