[radvd-devel-l] regression with version 1.8

Mark Smith radvd at 02a76c927861ca7413a122f2a73a0d37.nosense.org
Thu Jun 2 19:37:20 EDT 2011


Hi George,

On Thu, 2 Jun 2011 20:27:34 +0300
"George K." <kargig.lists at gmail.com> wrote:

> Due to my config (posted previously) the new function
> cease_adv_pfx_msg in send.c gets called and this function in turn
> calls print_addr.

I find that very strange. cease_adv_pfx_msg is only called if the
DecrementLifetimes prefix option is set, and the preferred lifetime of
the prefix reaches zero, yet your config doesn't have the per-prefix 
DecrementLifetimes flag set, and it doesn't default to on.

Are your lifetimes decrementing, even though the DecrementLifetimes
option isn't enabled? What is the behaviour if you do enable it?

> Then inet_ntop gets called and...messes up the
> variables (?) I know this seems irrational, but if I avoid bumping
> into print_addr by changing my config, radvd keeps working.
> 
> I did some tests today and if I advertise prefixes/rdnss from another
> host then proccess.c reads them fine and prints correct msgs (like
> those of line 338 in process.c). There print_addr is also used but it
> does not cause trouble as it does inside send.c.
> 
> If you need me to do any further testing, please let me know.
> 
> $ file radvd
> radvd: ELF 32-bit LSB executable, ARM, version 1, dynamically linked
> (uses shared libs), stripped
> -- 
> George Kargiotakis
> 
Regards,
Mark.



More information about the radvd-devel-l mailing list