Greetings,
I'm trying to compile a 2.6.2x kernel on a Linkstation/KuroBox HG/ppc
following instructions given under
http://www.kurobox.com/mwiki/index.php/Kurobox_support_in_stock_2.6_kernels
and get weired settings and errors from the serial device (UART 16550A).
Under the 2.4.17 kernel provided by the Kurobox manufacturer it gets
IRQS in the range of 137/138 under 2.6.22 for the linkstation it gets
IRQS 17/18 and the serial device behaves fishy
communication errors
"to much work in IRQ 16"
stty -F /dev/ttyS0 sometimes tells me
"unable to perform all requested operations"
setserial tells me the port for /dev/ttyS0 is 0x000 (?)
while cat /proc/tty/driver/serial reports the proper mmio
0x80004500
This holds true for 2.6.20 and 2.6.22.1
Any suggestions ?
Thanks for listening.
On Wed, 8 Aug 2007, Oncaphillis wrote:
> Greetings,
>
> I'm trying to compile a 2.6.2x kernel on a Linkstation/KuroBox HG/ppc
> following instructions given under
> http://www.kurobox.com/mwiki/index.php/Kurobox_support_in_stock_2.6_kernels
Wow, congratulations! Not too many endevour into this land, and even fewer
succeed:-) You'd probably be better off asking this on linuxppc-dev, but
anyway...
> and get weired settings and errors from the serial device (UART 16550A).
>
> Under the 2.4.17 kernel provided by the Kurobox manufacturer it gets IRQS in
> the range of 137/138 under 2.6.22 for the linkstation it gets IRQS 17/18 and
> the serial device behaves fishy
That interrupt numbers are different as under 2.4 is correct, however, I
get 16 and 17 for ttyS0 and ttyS1 on a kuroboxHG, are you sure you're
getting 17 and 18?
> communication errors
>
> "to much work in IRQ 16"
And here you get IRQ 16, which indeed should be ttyS0. And ttyS0 is
connected to the power-management microcontroller (AVR). Wait, what
userspace are you running? Have you swapped ttyS0 and ttyS1 everywhere? in
/etc/inittab for example? Besides, if you are running any vendor provided
(or community) tools to handle the AVR - they use ttyS1 by default, as it
was under 2.4 ans under "traditional" 2.6, whereas the 2.6 mainline kernel
for kurobox/linkstation uses the traditional for these CPUs port
numbering. I think this is your problem. Some user-space programme(s) are
accessing wrong tty.
Thanks
Guennadi
---
Guennadi Liakhovetski
Guennadi Liakhovetski wrote:
> On Wed, 8 Aug 2007, Oncaphillis wrote:
>
>> Greetings,
>>
>> I'm trying to compile a 2.6.2x kernel on a Linkstation/KuroBox HG/ppc
>> following instructions given under
>> http://www.kurobox.com/mwiki/index.php/Kurobox_support_in_stock_2.6_kernels
>
> Wow, congratulations! Not too many endevour into this land, and even fewer
> succeed:-) You'd probably be better off asking this on linuxppc-dev, but
> anyway...
>
Thank you very much indeed. I'm kind of proud too that I dared to enter
this dark realm without bricking the system (at least up until now).
And thanks for the tip. I'll give it a try.
>> and get weired settings and errors from the serial device (UART 16550A).
>>
>> Under the 2.4.17 kernel provided by the Kurobox manufacturer it gets IRQS in
>> the range of 137/138 under 2.6.22 for the linkstation it gets IRQS 17/18 and
>> the serial device behaves fishy
>
> That interrupt numbers are different as under 2.4 is correct, however, I
> get 16 and 17 for ttyS0 and ttyS1 on a kuroboxHG, are you sure you're
> getting 17 and 18?
>
Absolutely right (bad bad typo, have to control my trembling fingers)
It's 16/17.
So that's ok ? No misconfiguration ? And that setserial tells
me the port is 0x000 ? I thought the dts file which
comes with the linkstation kernel describes the ports and IRQs
but poking there didn't change anything. It seems like the kernel
is ignoring the serial settings in the dts-file (which I do not
realy understand) all together.
>> communication errors
>>
>> "to much work in IRQ 16"
>
> And here you get IRQ 16, which indeed should be ttyS0. And ttyS0 is
> connected to the power-management microcontroller (AVR). Wait, what
> userspace are you running? Have you swapped ttyS0 and ttyS1 everywhere? in
> /etc/inittab for example? Besides, if you are running any vendor provided
> (or community) tools to handle the AVR - they use ttyS1 by default, as it
> was under 2.4 ans under "traditional" 2.6, whereas the 2.6 mainline kernel
> for kurobox/linkstation uses the traditional for these CPUs port
> numbering. I think this is your problem. Some user-space programme(s) are
> accessing wrong tty.
I started with a debian distro
http://www.kurobox.com/mwiki/index.php/Debian_sylver
Hmm I tried my very best to eliminate this everywhere. Actually I
resurrected the ARV00 device by creating a soft link to a ttySx
with during bootup within the udev filesystem.
<snip>
if [ ! -e /dev/AVR00 ];then
cat /proc/tty/driver/serial | grep 80004500 | grep -q 0:
&& AVR=/dev/ttyS0 || AVR=/dev/ttyS1
echo "Making AVR00 link to "$AVR;
ln -s $AVR /dev/AVR00
fi;
</snip>
Stole the trick from the avr_evtd startup script...
Thanks a 10^6 for the tips...
Oncaphillis
>
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
On Wed, 8 Aug 2007, Oncaphillis wrote:
> So that's ok ? No misconfiguration ? And that setserial tells
> me the port is 0x000 ?
Yes, everything looks ok, don't know why, setserial reports 0 here too,
but it doesn't cause me any (visible) problems.
> I thought the dts file which
> comes with the linkstation kernel describes the ports and IRQs
> but poking there didn't change anything. It seems like the kernel
> is ignoring the serial settings in the dts-file (which I do not
> realy understand) all together.
Believe me, it doesn't ignore them:-) Don't know what exactly you were
poking at there, but it should be quite easy to break it by changing
clocks / address / irq...
> I started with a debian distro
> http://www.kurobox.com/mwiki/index.php/Debian_sylver
>
> Hmm I tried my very best to eliminate this everywhere. Actually I
> resurrected the ARV00 device by creating a soft link to a ttySx
> with during bootup within the udev filesystem.
>
> <snip>
> if [ ! -e /dev/AVR00 ];then
> cat /proc/tty/driver/serial | grep 80004500 | grep -q 0:
> && AVR=/dev/ttyS0 || AVR=/dev/ttyS1
> echo "Making AVR00 link to "$AVR;
> ln -s $AVR /dev/AVR00
> fi;
> </snip>
>
> Stole the trick from the avr_evtd startup script...
Well, just do "ls -l /dev/AVR00", then "lsof /dev/ttyS0".
In principle, as mentioned elsewhere before, you don't need any daemons on
linkstation/kurobox now for "basic" operation, i.e., the kernel disables
the watchdog on startup, and properly sends reboot and power-off commands
to the AVR itself without any userspace support. So, at least for a test,
you could stop all those daemons. You need them though if you want to use
the buttons to power down / reboot / whatever, do other fancy things with
the AVR...
Another possible reason why, for example, your stty misbehaves, maybe it
is somehow broken in that distro? I am just using etch and things work
pretty well.
Thanks
Guennadi
---
Guennadi Liakhovetski
Guennadi Liakhovetski wrote:
> On Wed, 8 Aug 2007, Oncaphillis wrote:
>
>> So that's ok ? No misconfiguration ? And that setserial tells
>> me the port is 0x000 ?
>
> Yes, everything looks ok, don't know why, setserial reports 0 here too,
> but it doesn't cause me any (visible) problems.
>
>> I thought the dts file which
>> comes with the linkstation kernel describes the ports and IRQs
>> but poking there didn't change anything. It seems like the kernel
>> is ignoring the serial settings in the dts-file (which I do not
>> realy understand) all together.
>
> Believe me, it doesn't ignore them:-) Don't know what exactly you were
> poking at there, but it should be quite easy to break it by changing
> clocks / address / irq...
>
>> I started with a debian distro
>> http://www.kurobox.com/mwiki/index.php/Debian_sylver
>>
>> Hmm I tried my very best to eliminate this everywhere. Actually I
>> resurrected the ARV00 device by creating a soft link to a ttySx
>> with during bootup within the udev filesystem.
>>
>> <snip>
>> if [ ! -e /dev/AVR00 ];then
>> cat /proc/tty/driver/serial | grep 80004500 | grep -q 0:
>> && AVR=/dev/ttyS0 || AVR=/dev/ttyS1
>> echo "Making AVR00 link to "$AVR;
>> ln -s $AVR /dev/AVR00
>> fi;
>> </snip>
>>
>> Stole the trick from the avr_evtd startup script...
>
> Well, just do "ls -l /dev/AVR00", then "lsof /dev/ttyS0".
>
Oh.... that's something I should have thought about (shaking my
head)
I hope I found the the misconfigured part. I didn't look
at /etc/inittab. It opens a getty on /dev/ttyS0. That
might do the trick. Have to recompile my kernel right now with
standard config to check this out.
> In principle, as mentioned elsewhere before, you don't need any daemons on
> linkstation/kurobox now for "basic" operation, i.e., the kernel disables
> the watchdog on startup, and properly sends reboot and power-off commands
> to the AVR itself without any userspace support. So, at least for a test,
> you could stop all those daemons. You need them though if you want to use
> the buttons to power down / reboot / whatever, do other fancy things with
> the AVR...
Well ... it doesn't seem to need the daemons but at least
the kuro_boot script or something similar has to be executed
once sending the right init strings to the AVR or your system
reboots after a few minutes at least on my box.
> Another possible reason why, for example, your stty misbehaves, maybe it
> is somehow broken in that distro? I am just using etch and things work
> pretty well.
>
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski
On Thu, 9 Aug 2007, Oncaphillis wrote:
>
> I hope I found the the misconfigured part. I didn't look
> at /etc/inittab. It opens a getty on /dev/ttyS0. That
> might do the trick. Have to recompile my kernel right now with
> standard config to check this out.
Hm, for that you don't have to recompile the kernel... unless you modified
it before, of course. Just change your inittab and reboot.
> Well ... it doesn't seem to need the daemons but at least
> the kuro_boot script or something similar has to be executed
> once sending the right init strings to the AVR or your system
> reboots after a few minutes at least on my box.
No, shouldn't be needed either. You should get a string in dmesg like
"linkstation: disarming the AVR watchdog:", and the watchdog should be
off. Just take the original .dts, unmodified kernel sources and close
enough to original (arch/powerpc/configs/linkstation_defconfig) .config
and it should work.
Thanks
Guennadi
---
Guennadi Liakhovetski
Guennadi Liakhovetski wrote:
> On Thu, 9 Aug 2007, Oncaphillis wrote:
>> I hope I found the the misconfigured part. I didn't look
>> at /etc/inittab. It opens a getty on /dev/ttyS0. That
>> might do the trick. Have to recompile my kernel right now with
>> standard config to check this out.
>
> Hm, for that you don't have to recompile the kernel... unless you modified
> it before, of course. Just change your inittab and reboot.
Oh I know... I just played around with the kernel config
to solve the problem and ended up with a kernel that didn't
boot at all..
>
>> Well ... it doesn't seem to need the daemons but at least
>> the kuro_boot script or something similar has to be executed
>> once sending the right init strings to the AVR or your system
>> reboots after a few minutes at least on my box.
>
> No, shouldn't be needed either. You should get a string in dmesg like
> "linkstation: disarming the AVR watchdog:", and the watchdog should be
> off. Just take the original .dts, unmodified kernel sources and close
> enough to original (arch/powerpc/configs/linkstation_defconfig) .config
> and it should work.
>
Thanks again...
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Removing the getty from initd did the trick.
And you're right AVR is disabled by the kernel
found the code in ./arch/powerpc/platforms/embedded6xx/ls_uart.c
Tkank you
Guennadi Liakhovetski wrote:
> On Thu, 9 Aug 2007, Oncaphillis wrote:
>> I hope I found the the misconfigured part. I didn't look
>> at /etc/inittab. It opens a getty on /dev/ttyS0. That
>> might do the trick. Have to recompile my kernel right now with
>> standard config to check this out.
>
> Hm, for that you don't have to recompile the kernel... unless you modified
> it before, of course. Just change your inittab and reboot.
>
>> Well ... it doesn't seem to need the daemons but at least
>> the kuro_boot script or something similar has to be executed
>> once sending the right init strings to the AVR or your system
>> reboots after a few minutes at least on my box.
>
> No, shouldn't be needed either. You should get a string in dmesg like
> "linkstation: disarming the AVR watchdog:", and the watchdog should be
> off. Just take the original .dts, unmodified kernel sources and close
> enough to original (arch/powerpc/configs/linkstation_defconfig) .config
> and it should work.
>
> Thanks
> Guennadi
> ---
> Guennadi Liakhovetski
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
On Thu, 9 Aug 2007, Oncaphillis wrote:
> Removing the getty from initd did the trick.
>
> And you're right AVR is disabled by the kernel
> found the code in ./arch/powerpc/platforms/embedded6xx/ls_uart.c
Good. For the future, with any (kernel / userspace / software / hardware /
weather / ...) linkstation/kurobox related questions you are welcome to
use [email protected] or the #linkstationwiki channel on
irc.freenode.net, or the forums on the http://linkstationwiki.net/.
Good luck
Guennadi
---
Guennadi Liakhovetski