[PATCH v1 net 0/2] fou/gue: Fix skb memleak with inner protocol 0.

[PATCH v1 net 0/2] fou/gue: Fix skb memleak with inner protocol 0.

[Kuniyuki Iwashima]

syzbot reported memleak for a GUE packet with its inner
protocol number 0.

Patch 1 fixes the issue, and patch 2 fixes the same issue
in FOU.


Kuniyuki Iwashima (2):
  gue: Fix skb memleak with inner IP protocol 0.
  fou: Don't allow 0 for FOU_ATTR_IPPROTO.

 net/ipv4/fou_core.c | 3 +++
 net/ipv4/fou_nl.c   | 2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

-- 
2.52.0.457.g6b5491de43-goog

[PATCH v1 net 1/2] gue: Fix skb memleak with inner IP protocol 0.

[Kuniyuki Iwashima]

syzbot reported skb memleak below. [0]

The repro generated a GUE packet with its inner protocol 0.

gue_udp_recv() returns -guehdr->proto_ctype for "resubmit"
in ip_protocol_deliver_rcu(), but this only works with
non-zero protocol number.

Let's drop such packets.

Note that 0 is a valid number (IPv6 Hop-by-Hop Option).

I think it is not practical to encap HOPOPT in GUE, so once
someone starts to complain, we could pass down a resubmit
flag pointer to distinguish two zeros from the upper layer:

  * no error
  * resubmit HOPOPT

[0]
BUG: memory leak
unreferenced object 0xffff888109695a00 (size 240):
  comm "syz.0.17", pid 6088, jiffies 4294943096
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
    00 40 c2 10 81 88 ff ff 00 00 00 00 00 00 00 00  .@..............
  backtrace (crc a84b336f):
    kmemleak_alloc_recursive include/linux/kmemleak.h:44 [inline]
    slab_post_alloc_hook mm/slub.c:4958 [inline]
    slab_alloc_node mm/slub.c:5263 [inline]
    kmem_cache_alloc_noprof+0x3b4/0x590 mm/slub.c:5270
    __build_skb+0x23/0x60 net/core/skbuff.c:474
    build_skb+0x20/0x190 net/core/skbuff.c:490
    __tun_build_skb drivers/net/tun.c:1541 [inline]
    tun_build_skb+0x4a1/0xa40 drivers/net/tun.c:1636
    tun_get_user+0xc12/0x2030 drivers/net/tun.c:1770
    tun_chr_write_iter+0x71/0x120 drivers/net/tun.c:1999
    new_sync_write fs/read_write.c:593 [inline]
    vfs_write+0x45d/0x710 fs/read_write.c:686
    ksys_write+0xa7/0x170 fs/read_write.c:738
    do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
    do_syscall_64+0xa4/0xf80 arch/x86/entry/syscall_64.c:94
    entry_SYSCALL_64_after_hwframe+0x77/0x7f

Fixes: 37dd0247797b1 ("gue: Receive side for Generic UDP Encapsulation")
Reported-by: syzbot+4d8c7d16b0e95c0d0f0d@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/netdev/6965534b.050a0220.38aacd.0001.GAE@google.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
 net/ipv4/fou_core.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/ipv4/fou_core.c b/net/ipv4/fou_core.c
index 3970b6b7ace53..ab8f309f8925d 100644
--- a/net/ipv4/fou_core.c
+++ b/net/ipv4/fou_core.c
@@ -215,6 +215,9 @@ static int gue_udp_recv(struct sock *sk, struct sk_buff *skb)
 		return gue_control_message(skb, guehdr);
 
 	proto_ctype = guehdr->proto_ctype;
+	if (unlikely(!proto_ctype))
+		goto drop;
+
 	__skb_pull(skb, sizeof(struct udphdr) + hdrlen);
 	skb_reset_transport_header(skb);
 
-- 
2.52.0.457.g6b5491de43-goog

[PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Kuniyuki Iwashima]

fou_udp_recv() has the same problem mentioned in the previous
patch.

If FOU_ATTR_IPPROTO is set to 0, skb is not freed by
fou_udp_recv() nor "resubmit"-ted in ip_protocol_deliver_rcu().

Let's forbid 0 for FOU_ATTR_IPPROTO.

Fixes: 23461551c0062 ("fou: Support for foo-over-udp RX path")
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
 net/ipv4/fou_nl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/net/ipv4/fou_nl.c b/net/ipv4/fou_nl.c
index 7a99639204b16..0dec9da1bff46 100644
--- a/net/ipv4/fou_nl.c
+++ b/net/ipv4/fou_nl.c
@@ -15,7 +15,7 @@
 const struct nla_policy fou_nl_policy[FOU_ATTR_IFINDEX + 1] = {
 	[FOU_ATTR_PORT] = { .type = NLA_BE16, },
 	[FOU_ATTR_AF] = { .type = NLA_U8, },
-	[FOU_ATTR_IPPROTO] = { .type = NLA_U8, },
+	[FOU_ATTR_IPPROTO] = { .type = NLA_U8, .min = 1 },
 	[FOU_ATTR_TYPE] = { .type = NLA_U8, },
 	[FOU_ATTR_REMCSUM_NOPARTIAL] = { .type = NLA_FLAG, },
 	[FOU_ATTR_LOCAL_V4] = { .type = NLA_U32, },
-- 
2.52.0.457.g6b5491de43-goog

Re: [PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Eric Dumazet]

On Mon, Jan 12, 2026 at 9:07 PM Kuniyuki Iwashima <kuniyu@google.com> wrote:
>
> fou_udp_recv() has the same problem mentioned in the previous
> patch.
>
> If FOU_ATTR_IPPROTO is set to 0, skb is not freed by
> fou_udp_recv() nor "resubmit"-ted in ip_protocol_deliver_rcu().
>
> Let's forbid 0 for FOU_ATTR_IPPROTO.
>
> Fixes: 23461551c0062 ("fou: Support for foo-over-udp RX path")
> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>

Reviewed-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH v1 net 1/2] gue: Fix skb memleak with inner IP protocol 0.

[Eric Dumazet]

On Mon, Jan 12, 2026 at 9:07 PM Kuniyuki Iwashima <kuniyu@google.com> wrote:
>
> syzbot reported skb memleak below. [0]
>
> The repro generated a GUE packet with its inner protocol 0.
>
> gue_udp_recv() returns -guehdr->proto_ctype for "resubmit"
> in ip_protocol_deliver_rcu(), but this only works with
> non-zero protocol number.
>
> Let's drop such packets.
>
> Note that 0 is a valid number (IPv6 Hop-by-Hop Option).
>
> I think it is not practical to encap HOPOPT in GUE, so once
> someone starts to complain, we could pass down a resubmit
> flag pointer to distinguish two zeros from the upper layer:
>
>   * no error
>   * resubmit HOPOPT

Reviewed-by: Eric Dumazet <edumazet@google.com>

Re: [PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Jakub Kicinski]

On Mon, 12 Jan 2026 20:06:36 +0000 Kuniyuki Iwashima wrote:
> fou_udp_recv() has the same problem mentioned in the previous
> patch.
> 
> If FOU_ATTR_IPPROTO is set to 0, skb is not freed by
> fou_udp_recv() nor "resubmit"-ted in ip_protocol_deliver_rcu().
> 
> Let's forbid 0 for FOU_ATTR_IPPROTO.
> 
> Fixes: 23461551c0062 ("fou: Support for foo-over-udp RX path")
> Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
> ---
>  net/ipv4/fou_nl.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/net/ipv4/fou_nl.c b/net/ipv4/fou_nl.c
> index 7a99639204b16..0dec9da1bff46 100644
> --- a/net/ipv4/fou_nl.c
> +++ b/net/ipv4/fou_nl.c
> @@ -15,7 +15,7 @@
>  const struct nla_policy fou_nl_policy[FOU_ATTR_IFINDEX + 1] = {
>  	[FOU_ATTR_PORT] = { .type = NLA_BE16, },
>  	[FOU_ATTR_AF] = { .type = NLA_U8, },
> -	[FOU_ATTR_IPPROTO] = { .type = NLA_U8, },
> +	[FOU_ATTR_IPPROTO] = { .type = NLA_U8, .min = 1 },
>  	[FOU_ATTR_TYPE] = { .type = NLA_U8, },
>  	[FOU_ATTR_REMCSUM_NOPARTIAL] = { .type = NLA_FLAG, },
>  	[FOU_ATTR_LOCAL_V4] = { .type = NLA_U32, },

This code is generated, please use :

diff --git a/Documentation/netlink/specs/fou.yaml b/Documentation/netlink/specs/fou.yaml
index 8e7974ec453f..331f1b342b3a 100644
--- a/Documentation/netlink/specs/fou.yaml
+++ b/Documentation/netlink/specs/fou.yaml
@@ -39,6 +39,8 @@ kernel-policy: global
       -
         name: ipproto
         type: u8
+        checks:
+          min: 1
       -
         name: type
         type: u8
diff --git a/net/ipv4/fou_nl.c b/net/ipv4/fou_nl.c
index 7a99639204b1..309d5ba983d0 100644
--- a/net/ipv4/fou_nl.c
+++ b/net/ipv4/fou_nl.c
@@ -15,7 +15,7 @@
 const struct nla_policy fou_nl_policy[FOU_ATTR_IFINDEX + 1] = {
 	[FOU_ATTR_PORT] = { .type = NLA_BE16, },
 	[FOU_ATTR_AF] = { .type = NLA_U8, },
-	[FOU_ATTR_IPPROTO] = { .type = NLA_U8, },
+	[FOU_ATTR_IPPROTO] = NLA_POLICY_MIN(NLA_U8, 1),
 	[FOU_ATTR_TYPE] = { .type = NLA_U8, },
 	[FOU_ATTR_REMCSUM_NOPARTIAL] = { .type = NLA_FLAG, },
 	[FOU_ATTR_LOCAL_V4] = { .type = NLA_U32, },

Re: [PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Kuniyuki Iwashima]

On Tue, Jan 13, 2026 at 7:11 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Mon, 12 Jan 2026 20:06:36 +0000 Kuniyuki Iwashima wrote:
> > fou_udp_recv() has the same problem mentioned in the previous
> > patch.
> >
> > If FOU_ATTR_IPPROTO is set to 0, skb is not freed by
> > fou_udp_recv() nor "resubmit"-ted in ip_protocol_deliver_rcu().
> >
> > Let's forbid 0 for FOU_ATTR_IPPROTO.
> >
> > Fixes: 23461551c0062 ("fou: Support for foo-over-udp RX path")
> > Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
> > ---
> >  net/ipv4/fou_nl.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/net/ipv4/fou_nl.c b/net/ipv4/fou_nl.c
> > index 7a99639204b16..0dec9da1bff46 100644
> > --- a/net/ipv4/fou_nl.c
> > +++ b/net/ipv4/fou_nl.c
> > @@ -15,7 +15,7 @@
> >  const struct nla_policy fou_nl_policy[FOU_ATTR_IFINDEX + 1] = {
> >       [FOU_ATTR_PORT] = { .type = NLA_BE16, },
> >       [FOU_ATTR_AF] = { .type = NLA_U8, },
> > -     [FOU_ATTR_IPPROTO] = { .type = NLA_U8, },
> > +     [FOU_ATTR_IPPROTO] = { .type = NLA_U8, .min = 1 },
> >       [FOU_ATTR_TYPE] = { .type = NLA_U8, },
> >       [FOU_ATTR_REMCSUM_NOPARTIAL] = { .type = NLA_FLAG, },
> >       [FOU_ATTR_LOCAL_V4] = { .type = NLA_U32, },
>
> This code is generated, please use :
>
> diff --git a/Documentation/netlink/specs/fou.yaml b/Documentation/netlink/specs/fou.yaml
> index 8e7974ec453f..331f1b342b3a 100644
> --- a/Documentation/netlink/specs/fou.yaml
> +++ b/Documentation/netlink/specs/fou.yaml
> @@ -39,6 +39,8 @@ kernel-policy: global
>        -
>          name: ipproto
>          type: u8
> +        checks:
> +          min: 1
>        -
>          name: type
>          type: u8
> diff --git a/net/ipv4/fou_nl.c b/net/ipv4/fou_nl.c
> index 7a99639204b1..309d5ba983d0 100644
> --- a/net/ipv4/fou_nl.c
> +++ b/net/ipv4/fou_nl.c
> @@ -15,7 +15,7 @@
>  const struct nla_policy fou_nl_policy[FOU_ATTR_IFINDEX + 1] = {
>         [FOU_ATTR_PORT] = { .type = NLA_BE16, },
>         [FOU_ATTR_AF] = { .type = NLA_U8, },
> -       [FOU_ATTR_IPPROTO] = { .type = NLA_U8, },
> +       [FOU_ATTR_IPPROTO] = NLA_POLICY_MIN(NLA_U8, 1),
>         [FOU_ATTR_TYPE] = { .type = NLA_U8, },
>         [FOU_ATTR_REMCSUM_NOPARTIAL] = { .type = NLA_FLAG, },
>         [FOU_ATTR_LOCAL_V4] = { .type = NLA_U32, },

Oh thanks, I missed it's auto-generated.

Btw I needed the change below to generate the diff above
by "./tools/net/ynl/ynl-regen.sh -f".  Maybe depending on


diff --git a/tools/net/ynl/ynl-regen.sh b/tools/net/ynl/ynl-regen.sh
index 81b4ecd891006..fda5fe24cfd47 100755
--- a/tools/net/ynl/ynl-regen.sh
+++ b/tools/net/ynl/ynl-regen.sh
@@ -29,9 +29,9 @@ for f in $files; do
  continue
     fi

-    echo -e "\tGEN ${params[2]}\t$f"
-    $TOOL --cmp-out --mode ${params[2]} --${params[3]} \
-   --spec $KDIR/${params[0]} $args -o $f
+    echo -e "\tGEN ${params[5]}\t$f"
+    $TOOL --cmp-out --mode ${params[4]} --${params[5]} \
+   --spec $KDIR/${params[1]} $args -o $f
 done

 popd >>/dev/null


fwiw, $params were like

3- Documentation/netlink/specs/fou.yaml
4: YNL-GEN kernel source
--
3- Documentation/netlink/specs/fou.yaml
4: YNL-GEN kernel header

Re: [PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Jakub Kicinski]

On Tue, 13 Jan 2026 23:15:42 -0800 Kuniyuki Iwashima wrote:
> Btw I needed the change below to generate the diff above
> by "./tools/net/ynl/ynl-regen.sh -f".  Maybe depending on
> 
> 
> diff --git a/tools/net/ynl/ynl-regen.sh b/tools/net/ynl/ynl-regen.sh
> index 81b4ecd891006..fda5fe24cfd47 100755
> --- a/tools/net/ynl/ynl-regen.sh
> +++ b/tools/net/ynl/ynl-regen.sh
> @@ -29,9 +29,9 @@ for f in $files; do
>   continue
>      fi
> 
> -    echo -e "\tGEN ${params[2]}\t$f"
> -    $TOOL --cmp-out --mode ${params[2]} --${params[3]} \
> -   --spec $KDIR/${params[0]} $args -o $f
> +    echo -e "\tGEN ${params[5]}\t$f"
> +    $TOOL --cmp-out --mode ${params[4]} --${params[5]} \
> +   --spec $KDIR/${params[1]} $args -o $f
>  done
> 
>  popd >>/dev/null
> 
> 
> fwiw, $params were like
> 
> 3- Documentation/netlink/specs/fou.yaml
> 4: YNL-GEN kernel source
> --
> 3- Documentation/netlink/specs/fou.yaml
> 4: YNL-GEN kernel header

Hm, I guess you have grep.lineNumber enabled in your git config?
Could you see if tossing --no-line-number into the git grep
in this script fixes it for you and if yes send a patch?

Re: [PATCH v1 net 2/2] fou: Don't allow 0 for FOU_ATTR_IPPROTO.

[Kuniyuki Iwashima]

On Wed, Jan 14, 2026 at 7:40 PM Jakub Kicinski <kuba@kernel.org> wrote:
>
> On Tue, 13 Jan 2026 23:15:42 -0800 Kuniyuki Iwashima wrote:
> > Btw I needed the change below to generate the diff above
> > by "./tools/net/ynl/ynl-regen.sh -f".  Maybe depending on
> >
> >
> > diff --git a/tools/net/ynl/ynl-regen.sh b/tools/net/ynl/ynl-regen.sh
> > index 81b4ecd891006..fda5fe24cfd47 100755
> > --- a/tools/net/ynl/ynl-regen.sh
> > +++ b/tools/net/ynl/ynl-regen.sh
> > @@ -29,9 +29,9 @@ for f in $files; do
> >   continue
> >      fi
> >
> > -    echo -e "\tGEN ${params[2]}\t$f"
> > -    $TOOL --cmp-out --mode ${params[2]} --${params[3]} \
> > -   --spec $KDIR/${params[0]} $args -o $f
> > +    echo -e "\tGEN ${params[5]}\t$f"
> > +    $TOOL --cmp-out --mode ${params[4]} --${params[5]} \
> > +   --spec $KDIR/${params[1]} $args -o $f
> >  done
> >
> >  popd >>/dev/null
> >
> >
> > fwiw, $params were like
> >
> > 3- Documentation/netlink/specs/fou.yaml
> > 4: YNL-GEN kernel source
> > --
> > 3- Documentation/netlink/specs/fou.yaml
> > 4: YNL-GEN kernel header
>
> Hm, I guess you have grep.lineNumber enabled in your git config?

Ah exactly, I didn't know that my corp machine enabled
it globally by default :)


> Could you see if tossing --no-line-number into the git grep
> in this script fixes it for you and if yes send a patch?

It worked, I'll include it in the series.

Thank you!