Hi Christian, are you heading for kernel inclusion? I browsed the mpc251x.c file and realized various issues which should be fixed for inclusion. How do you want to proceed? Fix SVN trunk first or directly prepare patches for inclusion (based on net-next-2.6) and do the backport thereafter?
christian pellegrin wrote: > Hi, > > On Mon, Oct 19, 2009 at 10:42 AM, Wolfgang Grandegger <[email protected]> > wrote: > >>> example I discovered by chance that when I disconnect the bus the >>> mcp2510 goes in rx error passive and when I short-circuit it goes into >>> tx error passive and then bus off). Any more testing is much >> That's the normal behavior, just the direction seems wrong. >> > > let me add that I was cangen-ing in both directions to catch race conditions. > > Here comes the patch: > > ---8<----- > This patch: > > - fixes a bug in interpreting RTR frames with no extended-id frames > > - fixes missing device name in bittiming struct and other minor > buglets. > > - implements correct behaviour on bus-off (according to > can_restart_ms setting). If can_restart_ms is zero the mcp251x > is put on sleep until restarted. A RESTARTED error messages should be sent in case of automatic recovery. I don't find it in the code. > - fixes not handling CANINTF_MERRF irq > > - takes out warnings that give false positives on bus errors > > - fixes sending multiple errors frames on bus state transitions. > Now only one frame is generated on each transition. > > Signed-off: Christian Pellegrin <[email protected]> For all other issues I prefer to comment on mpc251x.c. Wolfgang. _______________________________________________ Socketcan-core mailing list [email protected] https://lists.berlios.de/mailman/listinfo/socketcan-core
