Wolfgang Grandegger wrote:
on the Xenomai mailing list the topic "bus error flooding" popped up
again. Various users reported trouble due to high bus error rates and
bad impact on latencies. Some discussion is going on on how to avoid
such flooding. I have already implemented "on-demand" bus error
interrupts. Bus error interrupts are then only enabled when at least one
socket is listening on bus errors. But flooding can still occur and we
are thinking about a better way of downscaling or temporarily disabling
them. Socket-CAN currently restarts the controller after 200 bus errors.
My preferred solution for RT-Socket-CAN currently is to stop the CAN
controller after a kernel configurable amount of successive bus errors.
More clever ideas and comments are welcome?
What do you think about the following method?
depends on XENO_DRIVERS_CAN_SJA1000
int "Maximum number of successive bus errors"
range 0 255
CAN bus errors are very useful for analyzing electrical problems
but they can come at a very high rate resulting in interrupt
flooding with bad impact on system performance and real-time
behavior. This option, if greater than 0, will limit the amount
of successive bus error interrupts. If the limit is reached, an
error message with "can_id = CAN_ERR_BUSERR_FLOOD" is sent. The
bus error counter gets reset on restart of the device and on any
successful message transmission or reception. Be aware that bus
error interrupts are only enabled if at least one socket is
listening on bus errors.
I think it should make Sebastian and Jan happy, at least ;-).
To have some input, I have measured the bus error rate with the PEAK
PCAN PCI card on my Icecube MPC5200 eval-board doing rtcansend without
cable connected. Here are the results for the various baud-rates:
125 KB/s 1926 BEI/s
250 KB/s 3925 BEI/s
500 KB/s 7856 BEI/s
1000 KB/s 15700 BEI/s
The latency measured with "latency" from the testsuite reported an
increase of the latency with load from 67 to 95us almost independently
of the baud-rate. Sending messages with 8 byte payload from MSCAN to
SJA1000 on the same node as fast as possible increased the latency up to
103us. This measurement did not include delivery of messages to sockets
(actually no socket was listening).
Xenomai-core mailing list