Re: s390 && user_enable_single_step() (Was: odd utrace testing results on s390x)

[Oleg Nesterov <oleg@redhat.com>]

On 01/05, Martin Schwidefsky wrote:
>
> On Tue, 5 Jan 2010 16:36:33 +0100
> Oleg Nesterov <oleg@redhat.com> wrote:
>
> > For example, why do_signal() sets TIF_SINGLE_STEP? Why can't we do
> > 
> > 	--- a/arch/s390/kernel/signal.c
> > 	+++ b/arch/s390/kernel/signal.c
> > 	@@ -500,18 +500,10 @@ void do_signal(struct pt_regs *regs)
> > 					clear_thread_flag(TIF_RESTORE_SIGMASK);
> > 	 
> > 				/*
> > 	-			 * If we would have taken a single-step trap
> > 	-			 * for a normal instruction, act like we took
> > 	-			 * one for the handler setup.
> > 	-			 */
> > 	-			if (current->thread.per_info.single_step)
> > 	-				set_thread_flag(TIF_SINGLE_STEP);
> > 	-
> > 	-			/*
> > 				 * Let tracing know that we've done the handler setup.
> > 				 */
> > 				tracehook_signal_handler(signr, &info, &ka, regs,
> > 	-					 test_thread_flag(TIF_SINGLE_STEP));
> > 	+					current->thread.per_info.single_step);
> > 			}
> > 			return;
> > 		}
> > 
> > ?
>
> The reason why we set the TIF_SINGLE_STEP bit in do_signal is that we
> want to be able to stop the debugged program before the first
> instruction of the signal handler has been executed. The PER single
> step causes a trap after an instruction has been executed. That first
> instruction can do bad things to the arguments of the signal handler..

Yes, but afaics all we need is to pass the correct "int stepping" arg
to tracehook_signal_handler(). If it is true, the tracee does
ptrace_notify() before it returns to user-mode.

> > dmesg shows 799 lines of
> >
> > 	XXX: 2389 0
> >
> >
> > The kernel is 2.6.32.2 + utrace, but CONFIG_UTRACE is not set.
>
> With or without my bug fix ?

With, but please see another email.


I'll add clear_bit(TIF_SINGLE_STEP) into do_fork() path and re-test.

Oleg.