[Xenomai] Error code -19

[Xenomai] Error code -19

[vibnwis]

Hi all,
I am a newbie and I believe code -19 has been asked before.

My system is made of

arch = arm
board = PandaBoard ES
Kernel = 3.18.12
Patch  =
xenomai-3/kernel/cobalt/arch/arm/patches/ipipe-core-3.18.12-arm-4.patch

output of dmesg | grep -i xenomai:
[    1.733825] [Xenomai] scheduling class idle registered.
[    1.733856] [Xenomai] scheduling class rt registered.
[    1.734588] [Xenomai] init failed, code -19


I had tried to reconfigure the kernel in accordance with the
http://www.xenomai.org/documentation/xenomai-head/html/TROUBLESHOOTING/#ENODEV
but the same errors produced.

Any comments would be appreciated.

Thank you

SiowLim

Re: [Xenomai] Error code -19

[Lowell Gilbert]

vibnwis <vibnwis@gmail.com> writes:

> I am a newbie and I believe code -19 has been asked before.

That is just reporting ENODEV.

It doesn't tell you much, although it usually means a missing timer.

> My system is made of
>
> arch = arm
> board = PandaBoard ES
> Kernel = 3.18.12
> Patch  =
> xenomai-3/kernel/cobalt/arch/arm/patches/ipipe-core-3.18.12-arm-4.patch
>
> output of dmesg | grep -i xenomai:
> [    1.733825] [Xenomai] scheduling class idle registered.
> [    1.733856] [Xenomai] scheduling class rt registered.
> [    1.734588] [Xenomai] init failed, code -19

If I-pipe reports finding a suitable timer, it may just be having
problems registering it. Look for I-pipe and scheduler messages in your
boot log.

Your board has some kind of OMAP processor. It's a Cortex-A9 CPU core,
so it should work without much problem. The example at
http://xenomai.org/2014/09/porting-xenomai-dual-kernel-to-a-new-arm-soc/#The_Cortex-A9_case
is dealing with an extremely similar chip.

Re: [Xenomai] Error code -19

[vibnwis]

I tried my best to understand what is being described in

----------------------
The Cortex-A9 case.
Starting with Linux 3.8, this clock frequency may be passed through the
device tree if the smp_twd timer is registered through the device tree. If
the timer is registered statically, starting with Linux 3.3 the smp_twd
timer could get the frequency from a clock named “smp_twd”.
------------------------
Here are things I found in my LInux kernel source.

"smp_twd.c"

static void twd_timer_setup(void) {
struct clock_event_device *clk = __this_cpu_ptr(twd_evt);
----------
#if defined(CONFIG_IPIPE) && defined(CONFIG_SMP)
    printk(KERN_INFO "I-pipe, %lu.%03lu MHz timer\n",
           twd_timer_rate / 1000000,
           (twd_timer_rate % 1000000) / 1000);
    clk->ipipe_timer = __this_cpu_ptr(&twd_itimer);
    clk->ipipe_timer->irq = clk->irq;
    clk->ipipe_timer->ack = twd_ack;
    clk->ipipe_timer->min_delay_ticks = 0xf;
#endif
----------
}

".config"
CONFIG_HAVE_SMP=y
CONFIG_SMP=y          <---- Enabled
CONFIG_SMP_ON_UP=y

....

# CONFIG_ARM_PSCI is not set
CONFIG_ARCH_NR_GPIO=512
CONFIG_IPIPE=y          <---- Enabled
# CONFIG_IPIPE_LEGACY is not set


smp_twd.c
#if defined(CONFIG_IPIPE) && defined(CONFIG_SMP)
    printk(KERN_INFO "I-pipe, %lu.%03lu MHz timer\n",
           twd_timer_rate / 1000000,
           (twd_timer_rate % 1000000) / 1000);
    clk->ipipe_timer = __this_cpu_ptr(&twd_itimer);
    clk->ipipe_timer->irq = clk->irq;
    clk->ipipe_timer->ack = twd_ack;
    clk->ipipe_timer->min_delay_ticks = 0xf;
#endif

CONFIG_HAVE_SMP=y
CONFIG_SMP=y          <---- Enabled
CONFIG_SMP_ON_UP=y

....

# CONFIG_ARM_PSCI is not set
CONFIG_ARCH_NR_GPIO=512
CONFIG_IPIPE=y          <---- Enabled
# CONFIG_IPIPE_LEGACY is not set

twd_timer_setup(void) has been registered.

However, I could not the timer is passed through device tree
"omap4-panda-es.dts", but I found the following, in omap4.dtsi. omap4.dtsi
is "included" in omap4460,dtsi.

local-timer@48240600 {
        compatible = "arm,cortex-a9-twd-timer";
        clocks = <&mpu_periphclk>;
        reg = <0x48240600 0x20>;
        interrupts = <GIC_PPI 13 (GIC_CPU_MASK_RAW(3) |
IRQ_TYPE_LEVEL_HIGH)>;
    };

"omap4-panda-es.dts"
#include "omap4460.dtsi
#include "omap4-panda-common.dtsi"
...............

However, I don't quite understand it thoroughly, Hence, I can really tell
if these are ok.

Can anyone help what could possibly go wrong my my Xenomai, please?




On 17 September 2015 at 05:24, Lowell Gilbert <kludge@be-well.ilk.org>
wrote:

> vibnwis <vibnwis@gmail.com> writes:
>
> > I am a newbie and I believe code -19 has been asked before.
>
> That is just reporting ENODEV.
>
> It doesn't tell you much, although it usually means a missing timer.
>
> > My system is made of
> >
> > arch = arm
> > board = PandaBoard ES
> > Kernel = 3.18.12
> > Patch  =
> > xenomai-3/kernel/cobalt/arch/arm/patches/ipipe-core-3.18.12-arm-4.patch
> >
> > output of dmesg | grep -i xenomai:
> > [    1.733825] [Xenomai] scheduling class idle registered.
> > [    1.733856] [Xenomai] scheduling class rt registered.
> > [    1.734588] [Xenomai] init failed, code -19
>
> If I-pipe reports finding a suitable timer, it may just be having
> problems registering it. Look for I-pipe and scheduler messages in your
> boot log.
>
> Your board has some kind of OMAP processor. It's a Cortex-A9 CPU core,
> so it should work without much problem. The example at
>
> http://xenomai.org/2014/09/porting-xenomai-dual-kernel-to-a-new-arm-soc/#The_Cortex-A9_case
> is dealing with an extremely similar chip.
>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Fri, Sep 18, 2015 at 01:02:53AM +1200, vibnwis wrote:
> I tried my best to understand what is being described in

Please post the full kernel logs.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

Thank you for looking into my problems.

Please find attached the kern.log.

Thanks in advance.

On 18 September 2015 at 01:32, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Fri, Sep 18, 2015 at 01:02:53AM +1200, vibnwis wrote:
> > I tried my best to understand what is being described in
>
> Please post the full kernel logs.
>
> --
>                                             Gilles.
> https://click-hack.org
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: kern.log
Type: text/x-log
Size: 558404 bytes
Desc: not available
URL: <http://xenomai.org/pipermail/xenomai/attachments/20150918/db23ee1a/attachment.bin>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Fri, Sep 18, 2015 at 01:39:36PM +1200, vibnwis wrote:
> Thank you for looking into my problems.
> 
> Please find attached the kern.log.
> 
> Thanks in advance.
> 
> On 18 September 2015 at 01:32, Gilles Chanteperdrix <
> gilles.chanteperdrix@xenomai.org> wrote:
> 
> > On Fri, Sep 18, 2015 at 01:02:53AM +1200, vibnwis wrote:
> > > I tried my best to understand what is being described in
> >
> > Please post the full kernel logs.
> >
> > --
> >                                             Gilles.
> > https://click-hack.org
> >

> Jan  1 00:00:07 arm kernel: [    1.454040] [Xenomai] scheduling class idle registered.
> Jan  1 00:00:07 arm kernel: [    1.454040] [Xenomai] scheduling class rt registered.
> Jan  1 00:00:07 arm kernel: [    1.454315] I-pipe: high-resolution clock not working
> Jan  1 00:00:07 arm kernel: [    1.454498] [Xenomai] init failed, code -19

So, the problem is the clock, not the timer. On a cortex A9, the
I-pipe uses the global timer as clock. Is the global timer enabled
in the kernel configuration, and is it declared in the device tree?

Regards.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

I have checked in .config. However, I am not sure what string to search for
device tree. Would you please advise what string would it be for the clock?

In .config

CONFIG_ARM_GLOBAL_TIMER=y
CONFIG_CLKSRC_ARM_GLOBAL_TIMER_SCHED_CLOCK=y



On 18 September 2015 at 19:46, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Fri, Sep 18, 2015 at 01:39:36PM +1200, vibnwis wrote:
> > Thank you for looking into my problems.
> >
> > Please find attached the kern.log.
> >
> > Thanks in advance.
> >
> > On 18 September 2015 at 01:32, Gilles Chanteperdrix <
> > gilles.chanteperdrix@xenomai.org> wrote:
> >
> > > On Fri, Sep 18, 2015 at 01:02:53AM +1200, vibnwis wrote:
> > > > I tried my best to understand what is being described in
> > >
> > > Please post the full kernel logs.
> > >
> > > --
> > >                                             Gilles.
> > > https://click-hack.org
> > >
>
> > Jan  1 00:00:07 arm kernel: [    1.454040] [Xenomai] scheduling class
> idle registered.
> > Jan  1 00:00:07 arm kernel: [    1.454040] [Xenomai] scheduling class rt
> registered.
> > Jan  1 00:00:07 arm kernel: [    1.454315] I-pipe: high-resolution clock
> not working
> > Jan  1 00:00:07 arm kernel: [    1.454498] [Xenomai] init failed, code
> -19
>
> So, the problem is the clock, not the timer. On a cortex A9, the
> I-pipe uses the global timer as clock. Is the global timer enabled
> in the kernel configuration, and is it declared in the device tree?
>
> Regards.
>
> --
>                                             Gilles.
> https://click-hack.org
>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sat, Sep 19, 2015 at 12:24:38AM +1200, vibnwis wrote:
> I have checked in .config. However, I am not sure what string to search for
> device tree. Would you please advise what string would it be for the clock?
> 
> In .config
> 
> CONFIG_ARM_GLOBAL_TIMER=y

Yes, this looks like the right option. Is the global timer declared
in the device tree?

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

I had tried, I could not find anything associated with global_timer or
global. Hence, may I know what wording would it be?

On 19 September 2015 at 00:29, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Sat, Sep 19, 2015 at 12:24:38AM +1200, vibnwis wrote:
> > I have checked in .config. However, I am not sure what string to search
> for
> > device tree. Would you please advise what string would it be for the
> clock?
> >
> > In .config
> >
> > CONFIG_ARM_GLOBAL_TIMER=y
>
> Yes, this looks like the right option. Is the global timer declared
> in the device tree?
>
> --
>                                             Gilles.
> https://click-hack.org
>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sat, Sep 19, 2015 at 12:42:11AM +1200, vibnwis wrote:
> I had tried, I could not find anything associated with global_timer or
> global. Hence, may I know what wording would it be?

Last time I tested Xenomai on omap4, the
"arm,cortex-a9-global-timer" compatible string could be found in
arch/arm/boot/dts/omap4.dtsi

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

yes, found this in arch/arm/boot/dts/omap4.dtsi.

global_timer: timer@48240200 {
        compatible = "arm,cortex-a9-global-timer";
        reg = <0x48240200 0x20>;
        clocks = <&mpu_periphclk>;
    };


Everything seems to be in place right?


On 19 September 2015 at 00:44, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Sat, Sep 19, 2015 at 12:42:11AM +1200, vibnwis wrote:
> > I had tried, I could not find anything associated with global_timer or
> > global. Hence, may I know what wording would it be?
>
> Last time I tested Xenomai on omap4, the
> "arm,cortex-a9-global-timer" compatible string could be found in
> arch/arm/boot/dts/omap4.dtsi
>
> --
>                                             Gilles.
> https://click-hack.org
>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sat, Sep 19, 2015 at 12:49:10AM +1200, vibnwis wrote:
> yes, found this in arch/arm/boot/dts/omap4.dtsi.
> 
> global_timer: timer@48240200 {
>         compatible = "arm,cortex-a9-global-timer";
>         reg = <0x48240200 0x20>;
>         clocks = <&mpu_periphclk>;
>     };
> 
> 
> Everything seems to be in place right?

Well, not really. Normally, the device tree for the global timer
should contain an irq declaration. We do not need it on omap4430
because drivers/clocksource/arm_gloabl_timer.c is badly hacked, but
the hack does not work on omap4460. The patch sent in another mail,
gets the hack also working on omap4460. But this is still a bad
hack.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

Thanks for sharing. Just finished compiling. Testing it soon.

On 19 September 2015 at 01:00, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Sat, Sep 19, 2015 at 12:49:10AM +1200, vibnwis wrote:
> > yes, found this in arch/arm/boot/dts/omap4.dtsi.
> >
> > global_timer: timer@48240200 {
> >         compatible = "arm,cortex-a9-global-timer";
> >         reg = <0x48240200 0x20>;
> >         clocks = <&mpu_periphclk>;
> >     };
> >
> >
> > Everything seems to be in place right?
>
> Well, not really. Normally, the device tree for the global timer
> should contain an irq declaration. We do not need it on omap4430
> because drivers/clocksource/arm_gloabl_timer.c is badly hacked, but
> the hack does not work on omap4460. The patch sent in another mail,
> gets the hack also working on omap4460. But this is still a bad
> hack.
>
> --
>                                             Gilles.
> https://click-hack.org
>

Re: [Xenomai] Error code -19

[vibnwis]

Hi Gilles,
Thank you so much. Finally, I got a working xenomai. Many thanks for the
help. Would this change be checked in?

Thank you once again.

On 19 September 2015 at 01:04, vibnwis <vibnwis@gmail.com> wrote:

> Thanks for sharing. Just finished compiling. Testing it soon.
>
> On 19 September 2015 at 01:00, Gilles Chanteperdrix <
> gilles.chanteperdrix@xenomai.org> wrote:
>
>> On Sat, Sep 19, 2015 at 12:49:10AM +1200, vibnwis wrote:
>> > yes, found this in arch/arm/boot/dts/omap4.dtsi.
>> >
>> > global_timer: timer@48240200 {
>> >         compatible = "arm,cortex-a9-global-timer";
>> >         reg = <0x48240200 0x20>;
>> >         clocks = <&mpu_periphclk>;
>> >     };
>> >
>> >
>> > Everything seems to be in place right?
>>
>> Well, not really. Normally, the device tree for the global timer
>> should contain an irq declaration. We do not need it on omap4430
>> because drivers/clocksource/arm_gloabl_timer.c is badly hacked, but
>> the hack does not work on omap4460. The patch sent in another mail,
>> gets the hack also working on omap4460. But this is still a bad
>> hack.
>>
>> --
>>                                             Gilles.
>> https://click-hack.org
>>
>
>

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sat, Sep 19, 2015 at 01:16:17AM +1200, vibnwis wrote:
> Hi Gilles,
> Thank you so much. Finally, I got a working xenomai. Many thanks for the
> help. Would this change be checked in?

I think the right change for omap4460 is to provide the global timer
irq in the device tree. I have to read some documentation to find
what that irq number is.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sun, Sep 20, 2015 at 07:57:17PM +0200, Gilles Chanteperdrix wrote:
> On Sat, Sep 19, 2015 at 01:16:17AM +1200, vibnwis wrote:
> > Hi Gilles,
> > Thank you so much. Finally, I got a working xenomai. Many thanks for the
> > help. Would this change be checked in?
> 
> I think the right change for omap4460 is to provide the global timer
> irq in the device tree. I have to read some documentation to find
> what that irq number is.

Following other device tree files, it seems the global timer irq is
the PPI 11. So, could you try the following patch instead of the
previous patch?

diff --git a/arch/arm/boot/dts/omap4.dtsi b/arch/arm/boot/dts/omap4.dtsi
index 2cb8144..e51b5e8 100644
--- a/arch/arm/boot/dts/omap4.dtsi
+++ b/arch/arm/boot/dts/omap4.dtsi
@@ -71,6 +71,7 @@
 		compatible = "arm,cortex-a9-global-timer";
 		reg = <0x48240200 0x20>;
 		clocks = <&mpu_periphclk>;
+		interrupts = <GIC_PPI 11 (GIC_CPU_MASK_RAW(3) | IRQ_TYPE_LEVEL_HIGH)>;
 	};
 
 	/*


Regards.

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[Gilles Chanteperdrix]

On Sat, Sep 19, 2015 at 12:42:11AM +1200, vibnwis wrote:
> I had tried, I could not find anything associated with global_timer or
> global. Hence, may I know what wording would it be?

Ok, the problem is that that code was tested on omap4430 where the
global timer can not be used by the mainline code. Could you try the
following patch?

diff --git a/drivers/clocksource/arm_global_timer.c b/drivers/clocksource/arm_global_timer.c
index 04ca5d3..191422d 100644
--- a/drivers/clocksource/arm_global_timer.c
+++ b/drivers/clocksource/arm_global_timer.c
@@ -279,7 +279,7 @@ static void __init global_timer_of_register(struct device_node *np)
 		gt_ppi = irq_of_parse_and_map(np, 0);
 		if (!gt_ppi) {
 			pr_warn("global-timer: unable to parse irq\n");
-			return;
+			usable_timer = 0;
 		}
 	}

-- 
					    Gilles.
https://click-hack.org

Re: [Xenomai] Error code -19

[vibnwis]

Many thanks, recompiling it now. Let you know tomorrow.

On 19 September 2015 at 00:50, Gilles Chanteperdrix <
gilles.chanteperdrix@xenomai.org> wrote:

> On Sat, Sep 19, 2015 at 12:42:11AM +1200, vibnwis wrote:
> > I had tried, I could not find anything associated with global_timer or
> > global. Hence, may I know what wording would it be?
>
> Ok, the problem is that that code was tested on omap4430 where the
> global timer can not be used by the mainline code. Could you try the
> following patch?
>
> diff --git a/drivers/clocksource/arm_global_timer.c
> b/drivers/clocksource/arm_global_timer.c
> index 04ca5d3..191422d 100644
> --- a/drivers/clocksource/arm_global_timer.c
> +++ b/drivers/clocksource/arm_global_timer.c
> @@ -279,7 +279,7 @@ static void __init global_timer_of_register(struct
> device_node *np)
>                 gt_ppi = irq_of_parse_and_map(np, 0);
>                 if (!gt_ppi) {
>                         pr_warn("global-timer: unable to parse irq\n");
> -                       return;
> +                       usable_timer = 0;
>                 }
>         }
>
> --
>                                             Gilles.
> https://click-hack.org
>