Cyril Wallois wrote: > Hi Jan, > > I join with this email the patch, don't hesitate email me if it is not > correct. I solve my oops problem by modifying the driver_remove routine. I > think the driver is now operational, the source code seems to be OK too.
Looks ok. Some remarks nevertheless: dm9000_hash_table is broken. It disabled the IRQ line, which is no good idea already, but it even does this before it is even initialized. Moreover, this function does not need locking as long as it is only called during init. Just drop it for now, maybe leaving a comment for the case we reuse dm9000_hash_table during runtime one day. And then I'm a bit worried about the MMIO-based data transfer under IRQ lock. How fast is this, e.g. how many microseconds does it take to write and read a full-sized packet? Doesn't this thing support DMA? To make it easier for me to merge your patch, please include also the new files into it. Then submit according to this guideline (see recent patches on rtnet-devel for further examples): http://thread.gmane.org/gmane.linux.real-time.rtnet.devel/388 Thanks for contributing! Jan
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------------
_______________________________________________ RTnet-users mailing list RTnet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rtnet-users
