On 19.01.2023 13:05:02, Thomas.Kopp@xxxxxxxxxxxxx wrote:
> Are you interrested in analyzing what's going on at those times?
Sure.
> If so, what to do in the driver to stop operation after the message
> has been issued so I can stop the analyzer and sample the chip
> registers?
If you return with an error value, the driver will dump the registers
with devcoredump and disable the IRQs and timestamping:
diff --git a/drivers/net/can/spi/mcp251xfd/mcp251xfd-tef.c b/drivers/net/can/spi/mcp251xfd/mcp251xfd-tef.c
index 0c89e4f33976..8f821ecde69b 100644
--- a/drivers/net/can/spi/mcp251xfd/mcp251xfd-tef.c
+++ b/drivers/net/can/spi/mcp251xfd/mcp251xfd-tef.c
@@ -79,6 +79,8 @@ mcp251xfd_handle_tefif_recover(const struct mcp251xfd_priv *priv, const u32 seq)
"not empty" : "empty",
seq, priv->tef->tail, priv->tef->head, tx_ring->head);
+ return -EINVAL;
+
/* The Sequence Number in the TEF doesn't match our tef_tail. */
return -EAGAIN;
}
Copy
https://github.com/linux-can/can-utils/blob/master/mcp251xfd/99-devcoredump.rules
to /etc/udev/rules.d and
https://github.com/linux-can/can-utils/blob/master/mcp251xfd/devcoredump
to /usr/sbin/
You'll find a coredump in /var/log/
> Absolutely interested, yes. I see the not empty cases pretty often as
> well and they shouldn't be an issue. I still can't reproduce the
> buffer full though. I pretty much only look at the raw SPI
> communication when analyzing this. So whatever you can capture on the
> SPI traces for SPI6 would be interesting for me to look at. If
> possible it would be great to capture either CANL or CANH as well.
> Yes, a full dump of the chip registers and RAM help to correlate
> things but in the end it's the pure SPI traces that help find
> problematic timings.
>
> Not sure how much you can log at a time but maybe it's possible to
> check the dmesg log and then kill your testscript and set a GPIO on
> the pi to trigger your logic analyzer to stop acquiring data? That
> would work if the analyzer can keep a couple seconds in a Ringbuffer
> or something.
If you don't want the devcoredump (which reads all chip registers)
comment it out in
https://elixir.bootlin.com/linux/v6.1.7/source/drivers/net/can/spi/mcp251xfd/mcp251xfd-core.c#L1578
Marc
--
Pengutronix e.K. | Marc Kleine-Budde |
Embedded Linux | https://www.pengutronix.de |
Vertretung West/Dortmund | Phone: +49-231-2826-924 |
Amtsgericht Hildesheim, HRA 2686 | Fax: +49-5121-206917-5555 |
Attachment:
signature.asc
Description: PGP signature
