On 2012-01-26 13:56, Jan Kiszka wrote:
>>> To trigger the enforced task termination without leaving any broken
>>> states behind, there is one option: rt_task_spin. Surprisingly for me,
>>> it actually spins in the kernel, thus triggers the second level if
>>> waiting long enough. I wonder, though, if that behavior shouldn't be
>>> improved, ie. the spinning loop be closed in user space - which would
>>> take away that option again.
>>>
>>> Thoughts?
>>>
>>
>> Tick-based timing is going to be the problem for determining the
>> spinning delay, unless we expose it in the vdso on a per-skin basis,
>> which won't be pretty.
>
> I see. But we should possibly add some signal-pending || amok test to
> that kernel loop. That would also kill my test design, but it makes
> otherwise some sense I guess.
I'm thinking of a change like this:
diff --git a/ksrc/skins/native/syscall.c b/ksrc/skins/native/syscall.c
index acf0375..39204b4 100644
--- a/ksrc/skins/native/syscall.c
+++ b/ksrc/skins/native/syscall.c
@@ -1020,13 +1020,21 @@ static int __rt_timer_inquire(struct pt_regs *regs)
static int __rt_timer_spin(struct pt_regs *regs)
{
+ xnthread_t *thread = xnpod_current_thread();
+ struct task_struct *p = current;
+ RTIME etime;
RTIME ns;
if (__xn_safe_copy_from_user(&ns, (void __user *)__xn_reg_arg1(regs),
sizeof(ns)))
return -EFAULT;
- rt_timer_spin(ns);
+ etime = xnarch_get_cpu_tsc() + xnarch_ns_to_tsc(ns);
+ while ((SRTIME)(xnarch_get_cpu_tsc() - etime) < 0) {
+ if (signal_pending(p) || xnthread_amok_p(thread))
+ return -EINTR;
+ cpu_relax();
+ }
return 0;
}
Regarding testability of the second watchdog state: We could add a
syscall to sigtest_module e.g which has the current rt_timer_spin
semantics.
Jan
--
Siemens AG, Corporate Technology, CT T DE IT 1
Corporate Competence Center Embedded Linux
_______________________________________________
Xenomai-core mailing list
Xenomai-core@gna.org
https://mail.gna.org/listinfo/xenomai-core
