On Fri, Nov 07, 2014 at 09:32:35PM +0100, Gilles Chanteperdrix wrote:
> On Fri, Nov 07, 2014 at 03:24:05PM -0500, Lennart Sorensen wrote:
> > On Fri, Nov 07, 2014 at 07:43:30PM +0100, Philippe Gerum wrote:
> > > For this particular issue, Xenomai 2.x is deadly wrong. Creating,
> > > deleting, enabling or disabling an IRQ channel must be done from
> > > secondary mode. The syscall mode bits for these native API calls are
> > > wrong, should be as follows instead:
> > >
> > > diff --git a/ksrc/skins/native/syscall.c b/ksrc/skins/native/syscall.c
> > > index 950b092..dff7f4d 100644
> > > --- a/ksrc/skins/native/syscall.c
> > > +++ b/ksrc/skins/native/syscall.c
> > > @@ -4145,12 +4145,12 @@ static xnsysent_t __systab[] = {
> > > [__native_alarm_stop] = {&__rt_alarm_stop, __xn_exec_any},
> > > [__native_alarm_wait] = {&__rt_alarm_wait, __xn_exec_primary},
> > > [__native_alarm_inquire] = {&__rt_alarm_inquire, __xn_exec_any},
> > > - [__native_intr_create] = {&__rt_intr_create, __xn_exec_any},
> > > + [__native_intr_create] = {&__rt_intr_create, __xn_exec_lostage},
> > > [__native_intr_bind] = {&__rt_intr_bind, __xn_exec_conforming},
> > > - [__native_intr_delete] = {&__rt_intr_delete, __xn_exec_any},
> > > + [__native_intr_delete] = {&__rt_intr_delete, __xn_exec_lostage},
> > > [__native_intr_wait] = {&__rt_intr_wait, __xn_exec_primary},
> > > - [__native_intr_enable] = {&__rt_intr_enable, __xn_exec_any},
> > > - [__native_intr_disable] = {&__rt_intr_disable, __xn_exec_any},
> > > + [__native_intr_enable] = {&__rt_intr_enable, __xn_exec_lostage},
> > > + [__native_intr_disable] = {&__rt_intr_disable, __xn_exec_lostage},
> > > [__native_intr_inquire] = {&__rt_intr_inquire, __xn_exec_any},
> > > [__native_pipe_create] = {&__rt_pipe_create, __xn_exec_lostage},
> > > [__native_pipe_bind] = {&__rt_pipe_bind, __xn_exec_conforming},
> >
> > I get the impression this might also be important:
> >
> > Index: linux-3.12.31.rr1/arch/arm/mach-omap2/timer.c
> > ===================================================================
> > --- linux-3.12.31.rr1.orig/arch/arm/mach-omap2/timer.c 2014-11-07
> > 10:10:47.736200170 -0500
> > +++ linux-3.12.31.rr1/arch/arm/mach-omap2/timer.c 2014-11-07
> > 11:42:03.807512821 -0500
> > @@ -824,7 +824,8 @@
> > omap2_sync32k_clocksource_init(); \
> > if (cpu_is_omap34xx()) \
> > omap3_pic_muter_register(); \
> > - else if (cpu_is_omap44xx() || soc_is_omap54xx()) \
> > + else if (cpu_is_omap44xx() || soc_is_omap54xx() \
> > + || soc_is_dra7xx()) \
> > omap4_pic_muter_register(); \
> > }
> > #endif
> >
> > The dra7xx (am57xx) we are using is very much like the omap54xx, and
> > use the same gpio system, so this call probably should be getting done.
> >
> > Does that seem plausible?
>
> This call enables "PIC muting". "PIC muting" means disabling non
> real-time interrupts when in the Xenomai domain, i.e. when executing
> real-time code. The I-pipe can work without that (as it does on many
> platforms). So, if you suspect you have IRQ handling issues, you may
> prefer to disable it, you do not want to suspect it of being the
> cause of the issues you have.
Well enabling it made no difference. It wasn't there before.
The syscall.c change got rid of the complaint about calling linux services
from xenomai domain so that seems like an improvement.
I still get the warning about set_affinity not being supported (which
is true of course).
I still get the lockup, but I am not sure the realtime irq handler is
doing its job properly yet so still investigating that. At least the
ugly warning is gone now.
--
Len Sorensen
_______________________________________________
Xenomai mailing list
[email protected]
http://www.xenomai.org/mailman/listinfo/xenomai
