Am 2014-08-05 11:52, schrieb Marc Kleine-Budde:

On 08/04/2014 06:01 PM, Stefan Agner wrote:

quoted

Am 2014-08-04 16:27, schrieb Marc Kleine-Budde:

quoted

On 08/04/2014 03:43 PM, Stefan Agner wrote:
[...]

quoted

quoted

Thanks for the test, so far looks promising :) With this setup the other
CAN node repeats the CAN frame until it's ACKed. Because there is no
node with a compatible bitrate, there is no ACking CAN node.

Can you add a third CAN node to the network. The second and third node
should use the same bitrate, while your vf610 uses a different one. With
the new setup it should take more than one frame until the vf610 goes
into error warning and even more frames to error passive. This way we
can see it the error warning interrupt is connected or not. The error
counters should increase with each "wrong" bitrate frame it sees, you
can check with:

    ip -details link show can0

The output looks like this:

quoted

4: can0: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN mode DEFAULT qlen 10
    link/can
    can state ERROR-ACTIVE (berr-counter tx 0 rx 0) restart-ms 0

                              ^^^^^^^^^^^^^^^^^^^^^^

quoted

    bitrate 1000000 sample-point 0.750
    tq 125 prop-seg 2 phase-seg1 3 phase-seg2 2 sjw 1
    sja1000: tseg1 1..16 tseg2 1..8 sjw 1..4 brp 1..64 brp-inc 1
    clock 8000000

When one of the berr-counter crosses 96 (and stays below 128) a warning
interrupt should be generated.

Ok, created this setup, could successfully communicate with all three
nodes. I then set the Vybrid to half of the bitrate. When I send a frame
from Vybrid, the berr-counter tx immediately ends up at 128 and the
device is in ERROR-PASSIVE:

This is expected.

quoted

root at colibri-vf:~# ip -details link show can1
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0
    can state ERROR-ACTIVE (berr-counter tx 0 rx 0) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

            ^^^^^^^^

BTW: the can core has a really weird clock rate, have a look at the
datasheet if you manage to route a 24 MHz clock (or another multiple of
8MHz) to the flexcan core. I had a quick glance at the datasheet, if I
understand it correctly the Fast OSC clock runs with 24 MHz.

This pinmux is actually part of the IPs CTRL register (see CLKSRC).
Currently this is set unconditionally to peripheral clock, which is on
my device 83.3 MHz (500MHz/6). This would be a bit bigger change. Just
thinking how to implement this.

Sigh! I've missed the fact, that this bit is _inside_ the CAN core
(again). On the mx53 and imx6 this bit was obsolete and the clock was
selected outside of the CAN core.

quoted

Maybe we have to reference a third clock
"osc" and a device tree property fsl,can-clock-osc which one can choose
between "osc" and "per" what do you think?

Yes, but that's a different patch. I thought it would be as easy as on
the mx53, where you just had to reconfigure the clock tree.

Agreed.

quoted

quoted

quoted

root at colibri-vf:~# cansend can1 1F334455#1122334455667788
interface = can1, family = 29, type = 3, proto = 1
root at colibri-vf:~# [  818.886664] flexcan_irq, esr=00062242
[  818.890365] flexcan_irq, ctrl=1c3dac57

root at colibri-vf:~# ip -details link show can1
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0
    can state ERROR-PASSIVE (berr-counter tx 128 rx 0) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421


When I send the frames from another device on the bus, I can see the rx
count incrementing by one on each frame I send. As you expected, the
device changes to ERROR-WARNING when crossing the 96 frame boundary:

This is correct

Just found out when using too high bitrate (500000 vs 125000), the error
counter immediately goes up to 128 (or even beyond) and ends up in
ERROR-PASSIVE:

Maybe there is a off-by-one error in the flexcan core, so that a rx
error of 127 doesn't trigger an error passive interrupt.

quoted

# ip -details link show can1
[  292.164820] flexcan_get_berr_counter, esr=00000000
[  292.169715] flexcan_get_berr_counter, esr=00040190
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0
    can state ERROR-PASSIVE (berr-counter tx 0 rx 135) restart-ms 0
    bitrate 298811 sample-point 0.777
    tq 371 prop-seg 3 phase-seg1 3 phase-seg2 2 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

[...]

quoted

quoted

quoted

However, once reaching 128, I don't get another interrupt and the device
stays in ERROR-WARNING:

The contents of the esr reg would be interesting, especially the
FLT_CONF part.

quoted

root at vybridvf61-v11:~# [  222.221651] flexcan_irq, esr=0005050a
[  222.225349] flexcan_irq, ctrl=1c3dac57

quoted

ip -details link show can1
[  223.791082] flexcan_get_berr_counter, esr=00000000
[  223.796246] flexcan_get_berr_counter, esr=00040180
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen
10
    link/can  promiscuity 0
    can state ERROR-WARNING (berr-counter tx 0 rx 96) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

quoted

root at vybridvf61-v11:~# ip -details link show can1
[  243.571175] flexcan_get_berr_counter, esr=00000000
[  243.576343] flexcan_get_berr_counter, esr=00040582
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen
10
    link/can  promiscuity 0
    can state ERROR-WARNING (berr-counter tx 0 rx 104) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

...

root at vybridvf61-v11:~# ip -details link show can1
[  299.831192] flexcan_get_berr_counter, esr=00000000
[  299.836358] flexcan_get_berr_counter, esr=00040582
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0
    can state ERROR-WARNING (berr-counter tx 0 rx 126) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421
root at vybridvf61-v11:~# ip -details link show can1
[  302.411025] flexcan_get_berr_counter, esr=00000000
[  302.416195] flexcan_get_berr_counter, esr=00040592
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0
    can state ERROR-WARNING (berr-counter tx 0 rx 128) restart-ms 0
    bitrate 124990 sample-point 0.739
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

The FLTCONF is in Error Passive, however the stack did not received that
information. I still have the printk in the interrupt, however I did not
get an interrupt here (while I get one when it switched to
ERROR-WARNING... But it looks like the ERRINT is still pending...

According to the datasheet the ERRINT indicates a bit error (bit 10...15
is set), not the error passive state.

Marc

True, but fact is, I don't get a interrupt when the system switches from
ERROR-WARNING to ERROR-PASSIVE state. However, when I set berr-reporting
on, I get an interrupt (probably because of a change of bit 10...15),
but the state change is recognized:

ip -details link show can1
[ 2303.571656] flexcan_get_berr_counter, esr=00000000
[ 2303.576832] flexcan_get_berr_counter, esr=00040180
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0 
    can <BERR-REPORTING> state ERROR-WARNING (berr-counter tx 0 rx 127)
restart-ms 0 
    bitrate 124990 sample-point 0.739 
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421
root at vybridvf61-v11:~# [ 2307.023789] flexcan_irq, esr=0004051a
[ 2307.027490] flexcan_irq, ctrl=1c3dec57
ip -details link show can1
[ 2309.081840] flexcan_get_berr_counter, esr=00000000
[ 2309.087016] flexcan_get_berr_counter, esr=00040190
3: can1: <NOARP,UP,LOWER_UP,ECHO> mtu 16 qdisc pfifo_fast state UNKNOWN
mode DEFAULT group default qlen 10
    link/can  promiscuity 0 
    can <BERR-REPORTING> state ERROR-PASSIVE (berr-counter tx 0 rx 128)
restart-ms 0 
    bitrate 124990 sample-point 0.739 
    tq 347 prop-seg 8 phase-seg1 8 phase-seg2 6 sjw 1
    flexcan: tseg1 4..16 tseg2 2..8 sjw 1..4 brp 1..256 brp-inc 1
    clock 83368421

It looks like, the FLTCONF state change itself doesn't lead to a
interrupt when the state changes to ERROR-PASSIVE. IMHO, this is also
not explicitly stated in the Vybrid RM (46.3.8).

What is the solution here? Force enabling ERR interrupt? The
FLEXCAN_HAS_BROKEN_ERR_STATE flag would be appropriate, however, this
was meant to work around the missing WRN_INT, which seems not to be the
case here...

--
Stefan