In linux-2.5.45, the core networking code calls
__secpath_destroy via the static inline routine secpath_put in
include/net/xfrm.h. However, __secpath destroy is defined in
ipv4. So, I believe that compiling networking without ipv4
will result in a kernel that fails to link (haven't actually
tried it), and it also causes problems for anyone who has
tweaked ipv4 into a loadable module (which is my case; I posted
patches long ago and would be happy to post them again if there
is interest).
Here is a possible patch that creates a secpath_destroy_hook,
although I hope that a cleaner and safer solution can be found (safer
because hook variables if multiple modules save and restore the old
values of the hook variable in some order other than
last-in-first-out).
I'm littering linux-kernel with this patch also because I
think __secpath_destroy comes from ipsec and those maintainers might
not be on the netdev and linux-net lists.
--
Adam J. Richter __ ______________ 575 Oroville Road
[email protected] \ / Milpitas, California 95035
+1 408 309-6081 | g g d r a s i l United States of America
"Free Software For The Rest Of Us."
From: "Adam J. Richter" <[email protected]>
Date: Fri, 1 Nov 2002 03:45:00 -0800
In linux-2.5.45, the core networking code calls
__secpath_destroy via the static inline routine secpath_put in
include/net/xfrm.h.
Yes, we are fully aware of this. It will be fixed in due time,
please use CONFIG_INET=y kernels for the time being.