On Sat, 5 Feb 2005 17:46:43 +1100
Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx> wrote:
> This doesn't work because net/core/dst.c can only search based
> on dst->dev. For the split device case, dst->dev is set to
> loopback_dev while rt6i_idev is set to the real device.
Indeed. I didn't catch that.
> If we wanted to preserve the split device semantics, then we
> can create a local GC list in IPv6 so that it can search based
> on rt6i_idev as well as the other keys.
Ok, so this would entail changing each ipv6 dst_free() call
into one to ip6_dst_free(), which would:
It would mean that dst_run_gc() would need to have some callback
like dst->ops->gc_destroy() or similar, which would allow ipv6
to delete the dst from it's local garbage list.
> Alternatively we can
> remove the dst->dev == dev check in dst_dev_event and dst_ifdown
> and move that test down to the individual ifdown functions.
I think there is a hole in this idea.... maybe.
If the idea is to scan dst_garbage_list down in ipv6 specific code,
you can't do that since 'dst' objects from every pool in the kernel
get put onto the dst_garbage_list. It is generic.
They have no identity, so it's illegal to treat any member of that
list as an rt_entry, rt6_entry or any specific higher level dst