[Swan-dev] protostack "noklips" broken with libevent

Paul Wouters paul at nohats.ca
Thu Sep 17 20:15:19 EEST 2015


On Wed, 16 Sep 2015, Paul Wouters wrote:

> Program received signal SIGSEGV, Segmentation fault.
> 0x00005555555baa29 in init_kernel () at
> /source/programs/pluto/kernel.c:2436
> 2436		ev_fd = pluto_event_new(*kernel_ops->async_fdp, EV_READ | 
> EV_PERSIST,
>
> This happens when starting pluto without having run ipsec _stackmanager.
>
> *kernel_ops->async_fdp is NULL and pluto_event_new() does not like it.
>
> On a separate note, I do not think we should auto-detect "no_kernel". We
> should only allow it when explicitely selected. Then this error would
> also be a more simple "no kernel stack found".

This change has been made, which prevents any crashes of mis-detecting
the protostack.

It still crashes when using protostack=none (eg kernel_noklips.c)

Is there still any advantage in supporting the "no stack" option?

Paul


More information about the Swan-dev mailing list