[En-Nut-Discussion] Testcase For Crashing Nut/OS

[Henrik Maier]

Ole Reinhardt wrote:
> Hi,
> 
> I have noticed a problem that could possibly result from the same
> reason. I use the sja1000 driver on our m-can plattform (ATMega128, 32K
> RAM, SJA1000, I2C, I2C I/O expander and lots of relais, character LCD).
> 
> The SJA1000 driver uses NutEventPostFromIRQ as well.
> 
> We now noticed a strange problem that after some time of medium load
> working our boards suddenly stop working. I could not figure out the
> reason until know.
> 
> Has anybody notices anything similar? Any suggestions where to start
> searching the bug?

I had to face similar issues with serial comms and short time outs under 
heavy interrupt load. The UART routines are also using 
NutEventPostFromIRQ/NutEventPost. No issues showed up when time-outs 
were turned off.

I submitted a set of initial patches yesterday to Harald which fixed the 
lock-ups for my test case. I am sure he will either publish those or 
more comprehensive/generic ones after full investigation of the scenario.

Henrik
http://www.proconx.com


> 
> Best reagards,
> 
> Ole Reinhardt
>