[Ovmsdev] CAN errata fix regression?

[Craig Leres]

On 7/18/21 1:58 PM, Michael Balzer wrote:
> one of the ESP32CAN errata fixes seems to have caused a regression.
> 
> I now frequently see this log message:
> 
> esp32can: TxCallback: fatal error: TX buffer not available
> 
> Meaning there was a TX buffer free interrupt, but the TX buffer status 
> did not match that. I never saw this message before the errata fixes.
> 
> Do others see this message?

I don't see those but (a) my cars just do simple obd-ii style polling 
and not beating on can that much and (b) I'm not sure I'm setup to log 
that error.

I have sd logging enabled, my default logging level is warnings, and I 
have these componet levels set:

     events	info
     gsm-nmea	info
     location	verbose
     v-cadillacc2cts/v-chevroletc6corvette
		verbose

what should I change?

I downloaded server logs for the car I've driven in the last week and 
there are a few of these:

     2021-07-15 07:48:36.968 PDT E (928486918) can: can1: 
intr=-835016074 rxpkt=11131488 txpkt=3262 errflags=0x8040d9 rxerr=0 
txerr=107 rxinval=0 rxovr=14 txovr=7474 txdelay=1141 txfail=2415
wdgreset=0 errreset=0
     2021-07-15 07:48:36.968 PDT E (928486918) can: can1: 
intr=-835016069 rxpkt=11131488 txpkt=3262 errflags=0x8048d9 rxerr=0 
txerr=131 rxinval=0 rxovr=14 txovr=7474 txdelay=1141 txfail=2415
wdgreset=0 errreset=0
     2021-07-15 07:48:36.968 PDT E (928486918) can: can1: 
intr=-835016065 rxpkt=11131488 txpkt=3262 errflags=0x8048d9 rxerr=0 
txerr=131 rxinval=0 rxovr=14 txovr=7474 txdelay=1141 txfail=2416
wdgreset=0 errreset=0

I've been looking at the 3/4 and 4/4 of the espressif errata changes and 
they involve saving all chip registers, resetting the chip, and 
restoring the registers...

		Craig