[PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[Florian Westphal]

From: Eric Dumazet <edumazet@google.com>

I added skb_vlan_inet_prepare() helper in the cited commit, hinting
that we would need to use it more broadly.

syzbot confirmed this was the case in ip6_gre.

uninit-value in ip6table_mangle_hook+0x97d/0x9c0 net/ipv6/netfilter/ip6table_mangle.c:72
 ip6t_mangle_out net/ipv6/netfilter/ip6table_mangle.c:56 [inline]
 ip6table_mangle_hook+0x97d/0x9c0 net/ipv6/netfilter/ip6table_mangle.c:72
 nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
 nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
 nf_hook include/linux/netfilter.h:269 [inline]
 __ip6_local_out+0x5ac/0x640 net/ipv6/output_core.c:143
 ip6_local_out+0x4c/0x210 net/ipv6/output_core.c:153
 ip6tunnel_xmit+0x129/0x460 include/net/ip6_tunnel.h:161
 ip6_tnl_xmit+0x341a/0x3860 net/ipv6/ip6_tunnel.c:1281

Uninit was stored to memory at:
 ip6_tnl_xmit+0x34f7/0x3860 net/ipv6/ip6_tunnel.c:1277
 __gre6_xmit+0x14b9/0x1550 net/ipv6/ip6_gre.c:815
 ip6gre_xmit_ipv4 net/ipv6/ip6_gre.c:839 [inline]
 ip6gre_tunnel_xmit+0x18f7/0x2030 net/ipv6/ip6_gre.c:922

Uninit was created at:
 slab_post_alloc_hook mm/slub.c:4091 [inline]
 slab_alloc_node mm/slub.c:4134 [inline]
 __do_kmalloc_node mm/slub.c:4263 [inline]
 __kmalloc_node_track_caller_noprof+0x6c7/0xf90 mm/slub.c:4283
 kmalloc_reserve+0x23e/0x4a0 net/core/skbuff.c:609
 pskb_expand_head+0x226/0x1a60 net/core/skbuff.c:2275
 skb_realloc_headroom+0x140/0x2b0 net/core/skbuff.c:2355
 ip6_tnl_xmit+0x2106/0x3860 net/ipv6/ip6_tunnel.c:1227
 __gre6_xmit+0x14b9/0x1550 net/ipv6/ip6_gre.c:815
 ip6gre_xmit_ipv4 net/ipv6/ip6_gre.c:839 [inline]
 ip6gre_tunnel_xmit+0x18f7/0x2030 net/ipv6/ip6_gre.c:922

Fixes: d8a6213d70ac ("geneve: fix header validation in geneve[6]_xmit_skb")
Reported-by: syzbot+6023ea32e206eef7920a@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=6023ea32e206eef7920a
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Mazin Al Haddad <mazin@getstate.dev>
Signed-off-by: Florian Westphal <fw@strlen.de>
---
 v3: pass 'true' argument to skb_vlan_inet_prepare to not change network
 header offset.

 net/ipv6/ip6_gre.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index d19d86ed4376..9e214c355e6c 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -881,7 +881,7 @@ static netdev_tx_t ip6gre_tunnel_xmit(struct sk_buff *skb,
 	__be16 payload_protocol;
 	int ret;
 
-	if (!pskb_inet_may_pull(skb))
+	if (skb_vlan_inet_prepare(skb, true))
 		goto tx_err;
 
 	if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))
@@ -929,7 +929,7 @@ static netdev_tx_t ip6erspan_tunnel_xmit(struct sk_buff *skb,
 	__u32 mtu;
 	int nhoff;
 
-	if (!pskb_inet_may_pull(skb))
+	if (skb_vlan_inet_prepare(skb, true))
 		goto tx_err;
 
 	if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))
-- 
2.52.0

Re: [PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[Eric Dumazet]

On Mon, Jan 19, 2026 at 12:25 PM Florian Westphal <fw@strlen.de> wrote:
>
> From: Eric Dumazet <edumazet@google.com>
>
> I added skb_vlan_inet_prepare() helper in the cited commit, hinting
> that we would need to use it more broadly.
>
> syzbot confirmed this was the case in ip6_gre.
>
> uninit-value in ip6table_mangle_hook+0x97d/0x9c0 net/ipv6/netfilter/ip6table_mangle.c:72
>  ip6t_mangle_out net/ipv6/netfilter/ip6table_mangle.c:56 [inline]
>  ip6table_mangle_hook+0x97d/0x9c0 net/ipv6/netfilter/ip6table_mangle.c:72
>  nf_hook_entry_hookfn include/linux/netfilter.h:154 [inline]
>  nf_hook_slow+0xf4/0x400 net/netfilter/core.c:626
>  nf_hook include/linux/netfilter.h:269 [inline]
>  __ip6_local_out+0x5ac/0x640 net/ipv6/output_core.c:143
>  ip6_local_out+0x4c/0x210 net/ipv6/output_core.c:153
>  ip6tunnel_xmit+0x129/0x460 include/net/ip6_tunnel.h:161
>  ip6_tnl_xmit+0x341a/0x3860 net/ipv6/ip6_tunnel.c:1281
>
> Uninit was stored to memory at:
>  ip6_tnl_xmit+0x34f7/0x3860 net/ipv6/ip6_tunnel.c:1277
>  __gre6_xmit+0x14b9/0x1550 net/ipv6/ip6_gre.c:815
>  ip6gre_xmit_ipv4 net/ipv6/ip6_gre.c:839 [inline]
>  ip6gre_tunnel_xmit+0x18f7/0x2030 net/ipv6/ip6_gre.c:922
>
> Uninit was created at:
>  slab_post_alloc_hook mm/slub.c:4091 [inline]
>  slab_alloc_node mm/slub.c:4134 [inline]
>  __do_kmalloc_node mm/slub.c:4263 [inline]
>  __kmalloc_node_track_caller_noprof+0x6c7/0xf90 mm/slub.c:4283
>  kmalloc_reserve+0x23e/0x4a0 net/core/skbuff.c:609
>  pskb_expand_head+0x226/0x1a60 net/core/skbuff.c:2275
>  skb_realloc_headroom+0x140/0x2b0 net/core/skbuff.c:2355
>  ip6_tnl_xmit+0x2106/0x3860 net/ipv6/ip6_tunnel.c:1227
>  __gre6_xmit+0x14b9/0x1550 net/ipv6/ip6_gre.c:815
>  ip6gre_xmit_ipv4 net/ipv6/ip6_gre.c:839 [inline]
>  ip6gre_tunnel_xmit+0x18f7/0x2030 net/ipv6/ip6_gre.c:922
>
> Fixes: d8a6213d70ac ("geneve: fix header validation in geneve[6]_xmit_skb")
> Reported-by: syzbot+6023ea32e206eef7920a@syzkaller.appspotmail.com
> Closes: https://syzkaller.appspot.com/bug?extid=6023ea32e206eef7920a
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> Cc: Mazin Al Haddad <mazin@getstate.dev>
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---
>  v3: pass 'true' argument to skb_vlan_inet_prepare to not change network
>  header offset.

Ah right, this escaped my radar, thanks Florian !

Re: [PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[Jakub Kicinski]

On Mon, 19 Jan 2026 12:24:57 +0100 Florian Westphal wrote:
> From: Eric Dumazet <edumazet@google.com>
> 
> I added skb_vlan_inet_prepare() helper in the cited commit, hinting
> that we would need to use it more broadly.

I _think_ this makes GRE forwarding tests a bit unhappy:

https://netdev.bots.linux.dev/contest.html?branch=net-next-2026-01-19--12-00&executor=vmksft-forwarding&pw-n=0&pass=0
-- 
pw-bot: cr

Re: [PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[Eric Dumazet]

On Mon, Jan 19, 2026 at 6:06 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 19 Jan 2026 12:24:57 +0100 Florian Westphal wrote:
> > From: Eric Dumazet <edumazet@google.com>
> >
> > I added skb_vlan_inet_prepare() helper in the cited commit, hinting
> > that we would need to use it more broadly.
>
> I _think_ this makes GRE forwarding tests a bit unhappy:
>
> https://netdev.bots.linux.dev/contest.html?branch=net-next-2026-01-19--12-00&executor=vmksft-forwarding&pw-n=0&pass=0
> --

I was unsure about ip6erspan_tunnel_xmit() change, I think I started
full tests days ago but probably was distracted.

I had :

diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index d19d86ed43766bbc8ec052113be02ab231a5272c..9e214c355e6ce15fa828866ae20fa8fe321b4bf7
100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -881,7 +881,7 @@ static netdev_tx_t ip6gre_tunnel_xmit(struct sk_buff *skb,
        __be16 payload_protocol;
        int ret;

-       if (!pskb_inet_may_pull(skb))
+       if (skb_vlan_inet_prepare(skb, true))
                goto tx_err;

        if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))
@@ -929,7 +929,7 @@ static netdev_tx_t ip6erspan_tunnel_xmit(struct
sk_buff *skb,
        __u32 mtu;
        int nhoff;

-       if (!pskb_inet_may_pull(skb))
+       if (skb_vlan_inet_prepare(skb, false))
                goto tx_err;

        if (!ip6_tnl_xmit_ctl(t, &t->parms.laddr, &t->parms.raddr))

Re: [PATCH net v3] ip6_gre: use skb_vlan_inet_prepare() instead of pskb_inet_may_pull()

[Florian Westphal]

Eric Dumazet <edumazet@google.com> wrote:
> On Mon, Jan 19, 2026 at 6:06 PM Jakub Kicinski <kuba@kernel.org> wrote:
> >
> > On Mon, 19 Jan 2026 12:24:57 +0100 Florian Westphal wrote:
> > > From: Eric Dumazet <edumazet@google.com>
> > >
> > > I added skb_vlan_inet_prepare() helper in the cited commit, hinting
> > > that we would need to use it more broadly.
> >
> > I _think_ this makes GRE forwarding tests a bit unhappy:
> >
> > https://netdev.bots.linux.dev/contest.html?branch=net-next-2026-01-19--12-00&executor=vmksft-forwarding&pw-n=0&pass=0
> > --
> 
> I was unsure about ip6erspan_tunnel_xmit() change, I think I started
> full tests days ago but probably was distracted.
> 
> I had :
> 
> diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
> index d19d86ed43766bbc8ec052113be02ab231a5272c..9e214c355e6ce15fa828866ae20fa8fe321b4bf7
> 100644
> --- a/net/ipv6/ip6_gre.c
> +++ b/net/ipv6/ip6_gre.c
> @@ -881,7 +881,7 @@ static netdev_tx_t ip6gre_tunnel_xmit(struct sk_buff *skb,
>         __be16 payload_protocol;
>         int ret;
> 
> -       if (!pskb_inet_may_pull(skb))
> +       if (skb_vlan_inet_prepare(skb, true))
>                 goto tx_err;

It has to be either true or false depending on test case 8-/

gre_gso.sh needs this to be set to true, skbs don't have a mac
header: with "false": skb nhoff gets munged from 0 to 14.

But in mirror_gre.sh test case, skbs do have a mac header:
"true" munges nh offset from 14 to 0 and test fails.