On Fri, May 30, 2003 at 02:43:35AM +0800, Zhang Haitao wrote: > and dear Greg Lindahl: > you just pointed out the function skb_over_panic(), but there are > many reasons will lead that fault! It is not an obvious bug, but this is an important clue. If you then look at skb_put (in include/linux/skbuff.h) you will find that you have written beyond the end of a skbuf, or a stray pointer has damaged the size of the skbuf. I would next put some debugging code before every skb_put() call (there are only 2 in sbmac.c) to see if it will overflow, and printk something if it is. Any kernel bug that you can make happen repeatedly can be debugged in this fashion. -- greg
