Restart after 4 min of operation?

Restart after 4 min of operation?

[Bill Gatliff]

Guys:


Can someone give me any ideas as to why my Beagleboard-like platform
would spontaneously restart after precisely four minutes of operation? 
No OOPS, no nothing.  This is with 2.6.32-rc5.


Thanks!


b.g.

-- 
Bill Gatliff
bgat@billgatliff.com

Re: Restart after 4 min of operation?

[Gary Thomas]

On 11/02/2009 08:35 PM, Bill Gatliff wrote:
> Guys:
>
>
> Can someone give me any ideas as to why my Beagleboard-like platform
> would spontaneously restart after precisely four minutes of operation?
> No OOPS, no nothing.  This is with 2.6.32-rc5.
>

BeagleBoard like?  More details are probably required for much help.

How do you power it?  My BeagleBoard is powered via the OTG.  I've
seen it reset if I hang too much on the USB host port (exceed the
meager power budget)

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------

Re: Restart after 4 min of operation?

[Bill Gatliff]

Gary Thomas wrote:
> On 11/02/2009 08:35 PM, Bill Gatliff wrote:
>> Guys:
>>
>>
>> Can someone give me any ideas as to why my Beagleboard-like platform
>> would spontaneously restart after precisely four minutes of operation?
>> No OOPS, no nothing.  This is with 2.6.32-rc5.
>>
>
> BeagleBoard like?  More details are probably required for much help.
>
> How do you power it?  My BeagleBoard is powered via the OTG.  I've
> seen it reset if I hang too much on the USB host port (exceed the
> meager power budget)

It's a clone of a BeagleBoard, but with more and friendlier I/O.  CUS
package, and they use the TPS65023RSB power supply manager.  I'm not
ready to bury anyone in more details yet, since it's a client's hardware
and I feel like I'm imposing already...  :)

When I boot it to my kernel, which is based on 2.6.32-rc5, it comes up
fine and then for no apparent reason after four minutes goes through a
reset again.  I know it's approximately the same each time, because I
have uptime running in a loop so that I can conveniently see when it
dies.  The very same tree boots and runs properly on a true BeagleBoard
rev C.

What's even more wierd is that even when sitting at the u-boot prompt
before launching any kernel at all, the board does the same thing---
spontaneous reset after about four minutes.  The client provided me the
u-boot image, and it says it's a modified one from the OMAP3EVM--- it
even provides that machine id. 

The above behavior makes me suspect hardware, but read on before you agree.

Finally, the client provided me a kernel of unknown origin (it says it's
a 2.6.29-rc2) that appears to run fine---- no hiccup at 4 minutes.  Of
course, that kernel doesn't have the features they need.  I'm going over
it now to see if there's a magic bit somewhere that mine is missing, but
I'm at a bit of a loss to explain what I'm seeing right now.

Anyone have any ideas?  Bizzare.


(BTW, nice to hear from you again, Gary!  It's been a while.)


b.g.

-- 
Bill Gatliff
bgat@billgatliff.com

RE: Restart after 4 min of operation?

[Gadiyar, Anand]

Gary Thomas wrote:
> > On 11/02/2009 08:35 PM, Bill Gatliff wrote:
> >> Guys:
> >>
> >>
> >> Can someone give me any ideas as to why my Beagleboard-like platform
> >> would spontaneously restart after precisely four minutes of operation?
> >> No OOPS, no nothing.  This is with 2.6.32-rc5.
> >>
> >
> > BeagleBoard like?  More details are probably required for much help.
> >
> > How do you power it?  My BeagleBoard is powered via the OTG.  I've
> > seen it reset if I hang too much on the USB host port (exceed the
> > meager power budget)
> 
> It's a clone of a BeagleBoard, but with more and friendlier I/O.  CUS
> package, and they use the TPS65023RSB power supply manager.  I'm not
> ready to bury anyone in more details yet, since it's a client's hardware
> and I feel like I'm imposing already...  :)
> 
> When I boot it to my kernel, which is based on 2.6.32-rc5, it comes up
> fine and then for no apparent reason after four minutes goes through a
> reset again.  I know it's approximately the same each time, because I
> have uptime running in a loop so that I can conveniently see when it
> dies.  The very same tree boots and runs properly on a true BeagleBoard
> rev C.
> 
> What's even more wierd is that even when sitting at the u-boot prompt
> before launching any kernel at all, the board does the same thing---
> spontaneous reset after about four minutes.  The client provided me the
> u-boot image, and it says it's a modified one from the OMAP3EVM--- it
> even provides that machine id.
> 
> The above behavior makes me suspect hardware, but read on before you agree.
> 
> Finally, the client provided me a kernel of unknown origin (it says it's
> a 2.6.29-rc2) that appears to run fine---- no hiccup at 4 minutes.  Of
> course, that kernel doesn't have the features they need.  I'm going over
> it now to see if there's a magic bit somewhere that mine is missing, but
> I'm at a bit of a loss to explain what I'm seeing right now.
> 
> Anyone have any ideas?  Bizzare.
> 

I suspect a watchdog timer of some kind, but I haven't got a clue which one.

There is a register - PRM_RSTST (address 0x48307258) that you could
read and maybe it will tell you what was the source of the reset.

- Anand

> 
> (BTW, nice to hear from you again, Gary!  It's been a while.)
> 
> 
> b.g.
> 
> --
> Bill Gatliff
> bgat@billgatliff.com

Re: Restart after 4 min of operation?

[Bill Gatliff]

Gadiyar, Anand wrote:
> I suspect a watchdog timer of some kind, but I haven't got a clue which one.
>   

Turns out I had disabled watchdog timer support in my kernel. 
Re-enabling it seems to have addressed the problem--- the system has run
for more than 5 minutes now.  :)

Apparently, the default for the OMAP watchdog timer is "armed" at
reset?  Interesting.


b.g.

-- 
Bill Gatliff
bgat@billgatliff.com

RE: Restart after 4 min of operation?

[Gadiyar, Anand]

Bill Gatliff wrote:
> Gadiyar, Anand wrote:
> > I suspect a watchdog timer of some kind, but I haven't got a clue which one.
> >
> 
> Turns out I had disabled watchdog timer support in my kernel.
> Re-enabling it seems to have addressed the problem--- the system has run
> for more than 5 minutes now.  :)
> 
> Apparently, the default for the OMAP watchdog timer is "armed" at
> reset?  Interesting.
> 
> 

Just read a little on the MPU watchdog timer. It looks like the default values would
have caused a reset after ~10s. So if your board is rebooting after 4 minutes,
maybe the bootloader is turning it on?

I've left a board on in the bootloader for hours without seeing a reboot, so either
the bootloader I was using disables the watchdog, or it's not armed at reset.

- Anand

Re: Restart after 4 min of operation?

[Gary Thomas]

On 11/03/2009 11:37 AM, Bill Gatliff wrote:
> Gadiyar, Anand wrote:
>> I suspect a watchdog timer of some kind, but I haven't got a clue which one.
>>
>
> Turns out I had disabled watchdog timer support in my kernel.
> Re-enabling it seems to have addressed the problem--- the system has run
> for more than 5 minutes now.  :)
>
> Apparently, the default for the OMAP watchdog timer is "armed" at
> reset?  Interesting.

I'm pretty sure this is set up in U-Boot.

If I were you, I'd ask the customer for the U-Boot modifications
so you can check what they've done.

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------