Re: [RFC] Proposal for ptrace improvements

[Denys Vlasenko <vda.linux@googlemail.com>]

On Wed, Mar 2, 2011 at 8:37 AM, Tejun Heo <tj@kernel.org> wrote:
> On Wed, Mar 02, 2011 at 12:16:23AM +0100, Denys Vlasenko wrote:
>> Let's spell this out in detail. Please correct me if
>> I misunderstood your proposal:
>>
>> We have a stopped task under ptrace.
>> (More precisely: debugger got a WSTOPPED notification via waitpid.
>> Debugger decided to emulate the job control stop, therefore it
>> keeps tracee stopped, therefore it just waits on waitpid
>> without doing any PTRACE_CONTs).
>>
>> Another task sends SIGCONT to the tracee.
>>
>> Debugger gets waitpid notification of the
>> WSTOPPED, WSTOPSIG == SIGCONT form.
>
> I think WSTOPSIG should be SIGTRAP as the tracee left group stop and
> entered ptrace trap.

This would be, by my count, 13th kind of SIGTRAP use by ptrace.
Which makes multi-level if's in debuggers even more complex
and more error-prone.

Why not SIGCONT? This event is, after all, caused by SIGCONT.
It would be so much nicer to be able to detect it with single if()
in the debugger...


>> Debugger can check PTRACE_GETSIGINFO, which succeeds.
>> Debugger now knows it's a signal delivery notification.
>
> No, it's not a signal delivery notification.  It's a ptrace trap
> notification.  SIGCONT may not be delivered to this task.  Please
> remember that it's the emission of SIGCONT which ends a group stop,
> not delivery.

>From userspace POV it's really a kernel's implementation detail.


>> Debugger performs PTRACE_CONT(SIGCONT) - it injects the signal.
>> [Question: what if debugger doesn't? IOW: is it possible
>> for debugger to suppress SIGCONTs, or not?
>
> SIGCONT shouldn't be used here and wouldn't make any difference.
> We're not in signal delivery path.
>
>> IOW2: what should happen if debugger
>> (a) does not do any PTRACE_CONT at all? or
>
> The tracee stays stopped.
>
>> (b) does PTRACE_CONT(<other_sig>)? or
>> (c) does PTRACE_CONT(0)?
>
> See above.

This means that SIGCONT handler will be executed in the tracee
after debugger does PTRACE_CONT(<any_signo>) at this point.

Which makes SIGCONT special: debugger can suppress execution
of other signal handlers in tracee, but not SIGCONT handler.
Another special case. Can we avoid having it?


>> Debugger gets WCONTINUED waitpid notification.
>> [question: do we need this?]
>
> I don't think we need this.  The tracer needs all the stopped
> notifications but it doesn't need the continued notification because a
> tracee is never continued without the tracer saying so.

Yes, I think it's ok.

-- 
vda