ColdFire 5272 and interrupt latency

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Hello, All!
I'm experimenting with interrupt latency on a Motorola ColdFire 5272 and
the timing is not very satisfactory.  I have a device connected to an
external interrupt line #2 (or #3 - configurable with jumpers).  I have
put a few commands accessing chip select 3 in order to connect to those
lines and see the timing on a scope.  The interrupt is of the highest
priority and the ISR does not use any of the kernel services
(uC/OS-II) - it doesn't even comply with uC/OS-II format of interrupts
as there is nothing (short of reset) that can, in turn, interrupt it.
The time interval between the interrupt line going down and the ISR
marking its entrance with CS3 access varies between 1uS up to 10+uS,
even though there are only LEA -60(A7), A7 (adjusting the stack) and
MOVEM.L D0-D7/A0-A6, (A7) (saving the generic-purpose registers)
commands before accessing chip select 3.  There are no other interrupts
of that priority (6) in the whole system, and OS_ENTER_CRITICAL() macro
was rewritten to mask all interrupts up to level 5.  I would expect the
whole prologue to take less than 500 nano-seconds, given this setup.
What can be the reason for such a slow reaction?  Are there any
instructions on this processor that can take so much time?

Thanks in advance,

Re: ColdFire 5272 and interrupt latency

Quoted text here. Click to load it

Probably not much help since I don't know the processor but some suggested
areas to look:
- Interrrupts disabled?
- Cache miss?
- Cache invalidation?
- DMA activity?
- Traps, Mults, Divs?
- Jtag activity?
- Bus monitor timeout?
- MOVEM with large number of registers?

Good Luck.
Alf Katz

Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system ( ).
We've slightly trimmed the long signature. Click to see the full one.

Site Timeline