On 20 Jun 2008, at 5:50pm, Jürgen Keil wrote: > That first parameter is a big data structure that gets > dynamically allocated / freed. Since there is "no mapping", > the structure must have been freed, but some part of > the kernel is still doing function calls to xnb_copy_to_peer > passing a pointer to the freed memory block as first > argument.
Agreed. That's "not supposed to happen." Looking at the xnb_t that was provided, it does appear that the backend went through the normal close path (ring pointers are NULL, etc.). Given that, nothing should be calling up into xnbo :-( I'll try to reproduce the failure here. _______________________________________________ xen-discuss mailing list [email protected]
