non-maskable interrupt

[Ilya Alexeev]

Hello Joel,

Friday, December 6, 2002 you wrote:

JS> Ilya Alexeev wrote:
>> 
>> Hi all.
>> 
>> I am using rtems-ss-20020301 and gen68360 bsp.

JS> I think this is the problem that tracked and fixed as PR267.  It
JS> was a bug in m68k/cpu_asm.S where it was not incrementing
JS> _ISR_Nest_level
JS> which violates the Watchdog* routines assumptions about knowing
JS> when an ISR happened.

JS> THe ChangeLog entry says it was fixed in August.  So anything newer
JS> than that has the fix.  CHeck out CVSweb to see what the change was
JS> or grab a newer version.

Joel, I have applied patch from the PR267 to my cpu/m68k/cpu_asm.S.
But it doesn't help.
Problems with Level 7 interrupts persist.

Also I tried to use cpu_asm.S from the latest (1118) snapshot,
but it doesn't help too.

What else I can try?

>> My MC68360 board has external device which uses
>> non-maskable Level 7 Interrupt Autovector.
>> 
>> RTEMS documentation says that it is OK to use non-maskable ISRs,
>> but they are must not use system calls.
>> 
>> My ISR doen't use any system calls,
>> it just read/write a registers of external device,
>> but there are some problems then this ISR called too often.
>> This ISR spoils some RTEMS internal data structures.
>> 
>> I have add a number of test tasks for my application,
>> which are just call rtems_task_wake_after(1).
>> Then I call the ISR enough often,
>> my system hangs somewhere in RTEMS' _Watchdog* routines...
>> 
>> What I am doing wrong?
>> Is it possible to correctly use non-maskable ISR in RTEMS?
>> 
>> And one more thing - I am using standart call - rtems_interrupt_catch
>> for attaching non-maskable ISR.. Is it OK?
>> 
>> I will be grateful for any help.
>> 
>> Ilya V. Alexeev
>> ilya at continuum.ru
>> 
>> Continuum Plus JSC
>> Yaroslavl, Russia
>> Phone: (+7)0852 240200
>> Fax:   (+7)0852 359977




Ilya V. Alexeev
ilya at continuum.ru

Continuum Plus JSC
Yaroslavl, Russia
Phone: (+7)0852 240200
Fax:   (+7)0852 359977