That first warning says that something stayed on a list even though it
was released (the 0x6b thing is the kmalloc free poison pattern). And
the oops looks related to something similar.
Randy, is this one also related to that ipv6 percpu list corruption?
IOW, does it go away with
like one of your other reports did?
And David - I think we need that patch merged. The error case for when
the percpu list entry is free'd without unlinking is _so_ annoying
(random crashes in totally unrelated code) that I think we need to get
that one closed asap. Hmm?
On Wed, Nov 3, 2010 at 4:16 PM, Randy Dunlap <email@example.com> wrote: