> Looks like it is indeed nvidia's fault. It called atomic_clear_short() > with an invalid pointer in nv_alloc_pages(). You might be able to look > at nv_alloc_pages() to try and figure out the bug.
nv_alloc_pages never actually calls atomic_clear_short(), but it does call several functions that call vm_object functions in FreeBSD's kernel that eventually call atomic_clear_short(). For some reason those functions in between aren't in the backtrace though, and without that I can (and have) look through the code in the kernel to see how nv_alloc_pages can get to atomic_clear_short through vm calls, but I'm not sure that's too awefully helpful. Ken To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-hackers" in the body of the message
