cvs commit: src/sys/netgraph/bluetooth/l2cap ng_l2cap_misc.c

[Maksim Yevmenkin]

Pawel,

> +> first, let me thank you for looking into the problem. i just want to
> +> ask you a couple of questions. did you manage to reproduce the
> +> problem? does your patch fix it? i would like to see the code path
> +> that triggers this panic.
> 
> I was able to reproduce it by doing:
> 
>         # l2ping -f -a somehost -s 65531
>         ^C
>         # /etc/rc.bluetooth stop ubt0

great, thanks!

> +> in the future, i would really appreciate if you submit your patches
> +> for review before committing them.
> 
> Sure, sorry about that. I assumed noone is interested, because noone
> offered help to submitter.

actually, i was looking into this. unfortunately my emails did not
make it to the list because someone broke our local mail system here.
one of the smtp hosts we use does not have dns entry and my mail is
rejected :( also i was very busy with real work for the last two
weeks, so i did not have any spare time to look into this.

please take no offence, but i'd like to be involved when bluetooth
code is changed. while i appreciate your work here, i'd like to see a
slightly different patch. in particular, i do think think that
ng_l2cap_free_con() should call ng_l2cap_xxx_untimeout() routines. 
imo, ng_l2cap_free_con()  should panic if it detects that callout is
pending. also L2CAP_CON_AUTO_DISCON_TIMO and NG_L2CAP_CON_LP_TIMO are
mutually exclusive. both use the same callout handle, so you can not
have both at the same time.

having said all that, i think your patch should stay for now. please
do not MFC it just yet. i will look into this first thing tomorrow
(monday).

thanks for your work again!

max