Hi, On Mon, Nov 16, 2009 at 9:58 PM, Paul Thomas <[email protected]> wrote:
> 3. Now I can read/write registers. I don't even need to use my hacked > /proc/file to test this. This is the only time "ip link set can0 type > can bitrate 50000" shows a actual numbers on the CNF debug line > "mcp251x spi1.3: CNF: 0x09 0xbf 0x02". > if I remember well you told me that the oscillator frequency is 20 Mhz ( = Fosc). So from the CNFs: SJW = 1 x Tq BRP = 9 => Tq = 2 * (9 + 1) / Fosc = .000001 s PHSEG1 = 8 x Tq PRSEG = 8 x Tq PHSEG2 = 3 x Tq so the baud rate is = 1 / tbit = 1 / ( (1 + 8 + 8 + 3) * Tq) = 50000 bit/s I think this is all right. I don't see any big problems as far as your other devices have similar parameters. Just to double check what we've seen privately: 1) does loopback mode work okay? 2) when you connect the bus you see that the mcp2515 goes in TX passive mode? 3) can you try another, higher, baud rate? > The only way I've found to get back to this state is to unload & > reload the driver. I understand that you can only mess with some of > the mcp251x registers when CANCTRL is in configuration mode, but at > the very least on startup you shouldn't need the failed link up > command for it to work. And probably whenever the link is down in > should be configurable. No, the configuration mode must be explicitly entered: "The MCP2515 must be initialized before activation. This is only possible if the device is in the Configuration mode. Configuration mode is automatically selected after power-up, a reset or can be entered from any other mode by setting the CANTRL.REQOP bits to `100'." so the driver does a reset before programming the CNF registers. -- Christian Pellegrin, see http://www.evolware.org/chri/ "Real Programmers don't play tennis, or any other sport which requires you to change clothes. Mountain climbing is OK, and Real Programmers wear their climbing boots to work in case a mountain should suddenly spring up in the middle of the computer room." _______________________________________________ Socketcan-core mailing list [email protected] https://lists.berlios.de/mailman/listinfo/socketcan-core
