netdev - Re: [PATCH net-next] netlink: correct nlmsg size for multicast notifications

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <CADXeF1GvpMOyTHOYaE5v6w+4jpBKjnT=he3qNpehghRWY+hNHQ@mail.gmail.com>
Date: Fri, 20 Dec 2024 00:27:57 +0900
From: Yuyang Huang <yuyanghuang@...gle.com>
To: Eric Dumazet <edumazet@...gle.com>
Cc: "David S. Miller" <davem@...emloft.net>, Jakub Kicinski <kuba@...nel.org>, 
	Paolo Abeni <pabeni@...hat.com>, Simon Horman <horms@...nel.org>, David Ahern <dsahern@...nel.org>, 
	roopa@...ulusnetworks.com, jiri@...nulli.us, stephen@...workplumber.org, 
	jimictw@...gle.com, prohr@...gle.com, liuhangbin@...il.com, 
	nicolas.dichtel@...nd.com, andrew@...n.ch, pruddy@...tta.att-mail.com, 
	netdev@...r.kernel.org, Maciej Żenczykowski <maze@...gle.com>, 
	Lorenzo Colitti <lorenzo@...gle.com>
Subject: Re: [PATCH net-next] netlink: correct nlmsg size for multicast notifications

>Also GFP_ATOMIC should probably be replaced by GFP_KERNEL.

>All inet6_ifmcaddr_notify() callers are in process context, hold RTNL,
>thus can sleep under memory pressure.

I might not have enough background knowledge here, but why does
holding RTNL imply that the callers are in process context?

Link: https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/tree/net/ipv6/addrconf.c#n2241

In addrconf.c, the logic joins the solicited-node multicast group and
calls into mcast.c logic, which eventually triggers the notification.
I guess this code path is not in process context when the kernel does
SLAAC?

Thanks,
Yuyang

On Thu, Dec 19, 2024 at 11:31 PM Yuyang Huang <yuyanghuang@...gle.com> wrote:
>
> Hi Eric
>
> Thanks for the prompt review feedback. I will adjust all the comments
> in the v2 patch.
>
> >inet6_fill_ifmcaddr() got  an EXPORT_SYMBOL() for no good reason,
> please remove it.
>
> I  made the same mistake in Link:
> https://lore.kernel.org/netdev/20241218090057.76899-1-yuyanghuang@google.com/T/
>
> I will fix that patch as well.
>
> Thanks,
> Yuyang
>
> Thanks,
> Yuyang
>
>
> On Thu, Dec 19, 2024 at 10:47 PM Eric Dumazet <edumazet@...gle.com> wrote:
> >
> > On Thu, Dec 19, 2024 at 2:42 PM Eric Dumazet <edumazet@...gle.com> wrote:
> > >
> > > On Thu, Dec 19, 2024 at 2:27 PM Yuyang Huang <yuyanghuang@...gle.com> wrote:
> > > >
> > > > Corrected the netlink message size calculation for multicast group
> > > > join/leave notifications. The previous calculation did not account for
> > > > the inclusion of both IPv4/IPv6 addresses and ifa_cacheinfo in the
> > > > payload. This fix ensures that the allocated message size is
> > > > sufficient to hold all necessary information.
> > > >
> > > > Fixes: 2c2b61d2138f ("netlink: add IGMP/MLD join/leave notifications")
> > > > Cc: Maciej Żenczykowski <maze@...gle.com>
> > > > Cc: Lorenzo Colitti <lorenzo@...gle.com>
> > > > Signed-off-by: Yuyang Huang <yuyanghuang@...gle.com>
> > > > ---
> > > >  net/ipv4/igmp.c  | 4 +++-
> > > >  net/ipv6/mcast.c | 4 +++-
> > > >  2 files changed, 6 insertions(+), 2 deletions(-)
> > > >
> > > > diff --git a/net/ipv4/igmp.c b/net/ipv4/igmp.c
> > > > index 8a370ef37d3f..4e2f1497f320 100644
> > > > --- a/net/ipv4/igmp.c
> > > > +++ b/net/ipv4/igmp.c
> > > > @@ -1473,7 +1473,9 @@ static void inet_ifmcaddr_notify(struct net_device *dev,
> > > >         int err = -ENOMEM;
> > > >
> > > >         skb = nlmsg_new(NLMSG_ALIGN(sizeof(struct ifaddrmsg)) +
> > > > -                       nla_total_size(sizeof(__be32)), GFP_ATOMIC);
> > > > +                       nla_total_size(sizeof(__be32)) +
> > > > +                       nla_total_size(sizeof(struct ifa_cacheinfo)),
> > > > +                       GFP_ATOMIC);
> > > >         if (!skb)
> > > >                 goto error;
> > > >
> > > > diff --git a/net/ipv6/mcast.c b/net/ipv6/mcast.c
> > > > index 587831c148de..b7430f15d1fc 100644
> > > > --- a/net/ipv6/mcast.c
> > > > +++ b/net/ipv6/mcast.c
> > > > @@ -920,7 +920,9 @@ static void inet6_ifmcaddr_notify(struct net_device *dev,
> > > >         int err = -ENOMEM;
> > > >
> > > >         skb = nlmsg_new(NLMSG_ALIGN(sizeof(struct ifaddrmsg)) +
> > > > -                       nla_total_size(16), GFP_ATOMIC);
> > > > +                       nla_total_size(16) +
> > >
> > > While we are at it , can you use nla_total_size(sizeof(struct in6_addr))
> > >
> > > inet6_fill_ifmcaddr() got  an EXPORT_SYMBOL() for no good reason,
> > > please remove it.
> > > Squash the following in v2, thanks !
> > >
> > > diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
> > > index 2e2684886953d55140cd3d4a1e024b5218331a49..4da409bc45777f60fd37bdee541c61165a51d22c
> > > 100644
> > > --- a/net/ipv6/addrconf.c
> > > +++ b/net/ipv6/addrconf.c
> > > @@ -5239,7 +5239,6 @@ int inet6_fill_ifmcaddr(struct sk_buff *skb,
> > >         nlmsg_end(skb, nlh);
> > >         return 0;
> > >  }
> > > -EXPORT_SYMBOL(inet6_fill_ifmcaddr);
> > >
> > >  static int inet6_fill_ifacaddr(struct sk_buff *skb,
> > >                                const struct ifacaddr6 *ifaca,
> >
> > Also GFP_ATOMIC should probably be replaced by GFP_KERNEL.
> >
> > All inet6_ifmcaddr_notify() callers are in process context, hold RTNL,
> > thus can sleep under memory pressure.
> >
> > Same remark for inet_ifmcaddr_notify()