---------- Forwarded message ---------- From: Lei Luo <[email protected]> Date: 2010/4/8 Subject: Re: [Socketcan-users] problem in cross compiling socketcan To: Kurt Van Dijck <[email protected]>
2010/4/8 Kurt Van Dijck <[email protected]> On Wed, Apr 07, 2010 at 07:18:34PM +0200, Oliver Hartkopp wrote: > > Lei Luo wrote: > > > > > > > > Another problem is: if I build socketcan as modules and loaded them > > > manually, an err will happen in sometime; if build it direct into > > > kernel, then load application modules such as vcan, things will be > > > OK. An err info is as follows(not every time the same): > > > > > > kernel BUG at mm/slab.c:602! > > > Unable to handle kernel NULL pointer dereference at virtual address > 00000000 > > > fcse pid: 0, 0x00000000 > > > pgd = c0004000 > > > [00000000] *pgd=00000000 > > > Internal error: Oops: 817 [#1] > > > Modules linked in: [last unloaded: can_dev] > > > > > > > Hm - no real idea. > > > > I discovered some problems if you build the kernel with CAN support and > then > > mix the modules at load time (e.g. can.ko from /lib/modules/... and > can.ko > > from the SVN tree). > > > > So when you e.g. load can-raw.ko from the SVN tree build you should also > load > > can.ko from the SVN tree before. > Ack. > To prevent the kernel CAN modules to load, I don't even compile the > kernel CAN support. This way, I'm sure I'm using SVN modules. > > That is a good idea. I do this at first, but then I patched SVN souces files into kernel manully, becaused I do not kown the function of each module and the dependents of them. Regards, Luo
_______________________________________________ Socketcan-users mailing list [email protected] https://lists.berlios.de/mailman/listinfo/socketcan-users
