On Tue, Dec 20, 2016 at 09:42:00PM +0100, Martin Pieuchot wrote:
> On 20/12/16(Tue) 20:01, Alexander Bluhm wrote:
> > On Tue, Dec 20, 2016 at 12:14:22PM +0100, Martin Pieuchot wrote:
> > > mld6_{start,stop}_listeting() are called in ioctl(2) context, thus at
> > > IPL_SOFTNET.
> >
> > These are
On 20/12/16(Tue) 20:01, Alexander Bluhm wrote:
> On Tue, Dec 20, 2016 at 12:14:22PM +0100, Martin Pieuchot wrote:
> > mld6_{start,stop}_listeting() are called in ioctl(2) context, thus at
> > IPL_SOFTNET.
>
> These are called from in6_addmulti() and in6_delmulti(). They still
> have a
On Tue, Dec 20, 2016 at 12:14:22PM +0100, Martin Pieuchot wrote:
> mld6_{start,stop}_listeting() are called in ioctl(2) context, thus at
> IPL_SOFTNET.
These are called from in6_addmulti() and in6_delmulti(). They still
have a splsoftnet() in them. Please remove this first and put an
assert
mld6_{start,stop}_listeting() are called in ioctl(2) context, thus at
IPL_SOFTNET.
mld6_fasttimeo(), like all *fasttimeo() routines is now always called
at IPL_SOFTNET.
So let's get rid of these recursive splsoftnet(), ok?
Index: netinet6/mld6.c