On 24/08/15(Mon) 20:32, Mike Belopuhov wrote:
> On Thu, Aug 20, 2015 at 20:10 +0200, Martin Pieuchot wrote:
> > On 20/08/15(Thu) 18:20, Mike Belopuhov wrote:
> [...]
> Thanks a lot for taking your time to dig through this.
> Your diff looks good to me apart from the small bit where
> you've moved
On 24/08/15(Mon) 22:40, Alexander Bluhm wrote:
> On Thu, Aug 20, 2015 at 01:37:00PM +0200, Martin Pieuchot wrote:
> > --- net/route.c 19 Aug 2015 13:27:38 - 1.223
> > +++ net/route.c 20 Aug 2015 11:25:54 -
> > @@ -570,12 +570,8 @@ rtdeletemsg(struct rtentry *rt, u_int ta
> >
On Thu, Aug 20, 2015 at 01:37:00PM +0200, Martin Pieuchot wrote:
> --- net/route.c 19 Aug 2015 13:27:38 - 1.223
> +++ net/route.c 20 Aug 2015 11:25:54 -
> @@ -570,12 +570,8 @@ rtdeletemsg(struct rtentry *rt, u_int ta
> info.rti_flags = rt->rt_flags;
> ifp = rt->
On Thu, Aug 20, 2015 at 20:10 +0200, Martin Pieuchot wrote:
> On 20/08/15(Thu) 18:20, Mike Belopuhov wrote:
> > Makes you wonder why the heck it wasn't done in the first place,
> > doesn't it?
>
> If you look at the original CSRG source tree, you'll see how/why
> this happened :)
>
> When karels@
On 20/08/15(Thu) 18:20, Mike Belopuhov wrote:
> Makes you wonder why the heck it wasn't done in the first place,
> doesn't it?
If you look at the original CSRG source tree, you'll see how/why
this happened :)
When karels@ changed rtrequest() to turn it into a frontend for the
radix tree in r7.8 o
Makes you wonder why the heck it wasn't done in the first place, doesn't it?
This is the next logical step after bluhm@'s retrequest1(9) unification.
The idea is to always return a route with an incremented reference count
when the ``rtp'' argument of rtrequest1(9) is non-NULL. Apart from the
code simplification this will prevent ``rt'' to be freed by another CPU
between
