Re: pch_can probable bug

[Wolfgang Grandegger <wg@grandegger.com>]

Hi Christian,

On 12/11/2012 05:08 PM, Christian Bendele wrote:
> Well, I just found the most recent patch to the generic c_can driver on this
> list. I had only checked with Kernel 3.7rcSomething before and it didn't seem to
> support the intel eg20t. I failed to check if there was a more recent patch
> available and if that supported the intel chip. I'm really sorry for that.
> 
> I will try if that patch works for me tomorrow, since unfortunatelly my working
> day is almost at its end and I don't have the hardware at home. However, cursery
> skimming the code of c_can.c, c_can_do_rx_poll() it seems to me as if it does
> the same thing here:
> 
> 
> 			if (msg_ctrl_save & IF_MCONT_EOB)
> 				return num_rx_pkts;

Just a quick answer. We know that the pch_can and also the c_can driver
does have issues with tx and rx. Races, out-of-order-receptions, etc. I
have sent patches for the C_CAN drivers to support the PCH as well. But
still the RX handling needs to be improved to avoid out-of-order
reception at high message input rate. I will tell more later today.

Wolfgang.