2021-04-05 18:10:37

by Phillip Potter

[permalink] [raw]
Subject: [PATCH] net: tun: set tun->dev->addr_len during TUNSETLINK processing

When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
to match the appropriate type, using new tun_get_addr_len utility function
which returns appropriate address length for given type. Fixes a
KMSAN-found uninit-value bug reported by syzbot at:
https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51

Reported-by: [email protected]
Signed-off-by: Phillip Potter <[email protected]>
---
drivers/net/tun.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++
1 file changed, 48 insertions(+)

diff --git a/drivers/net/tun.c b/drivers/net/tun.c
index 978ac0981d16..56c26339ee3b 100644
--- a/drivers/net/tun.c
+++ b/drivers/net/tun.c
@@ -69,6 +69,14 @@
#include <linux/bpf.h>
#include <linux/bpf_trace.h>
#include <linux/mutex.h>
+#include <linux/ieee802154.h>
+#include <linux/if_ltalk.h>
+#include <uapi/linux/if_fddi.h>
+#include <uapi/linux/if_hippi.h>
+#include <uapi/linux/if_fc.h>
+#include <net/ax25.h>
+#include <net/rose.h>
+#include <net/6lowpan.h>

#include <linux/uaccess.h>
#include <linux/proc_fs.h>
@@ -2925,6 +2933,45 @@ static int tun_set_ebpf(struct tun_struct *tun, struct tun_prog __rcu **prog_p,
return __tun_set_ebpf(tun, prog_p, prog);
}

+/* Return correct value for tun->dev->addr_len based on tun->dev->type. */
+static inline unsigned char tun_get_addr_len(unsigned short type)
+{
+ switch (type) {
+ case ARPHRD_IP6GRE:
+ case ARPHRD_TUNNEL6:
+ return sizeof(struct in6_addr);
+ case ARPHRD_IPGRE:
+ case ARPHRD_TUNNEL:
+ case ARPHRD_SIT:
+ return 4;
+ case ARPHRD_ETHER:
+ return ETH_ALEN;
+ case ARPHRD_IEEE802154:
+ case ARPHRD_IEEE802154_MONITOR:
+ return IEEE802154_EXTENDED_ADDR_LEN;
+ case ARPHRD_PHONET_PIPE:
+ case ARPHRD_PPP:
+ case ARPHRD_NONE:
+ return 0;
+ case ARPHRD_6LOWPAN:
+ return EUI64_ADDR_LEN;
+ case ARPHRD_FDDI:
+ return FDDI_K_ALEN;
+ case ARPHRD_HIPPI:
+ return HIPPI_ALEN;
+ case ARPHRD_IEEE802:
+ return FC_ALEN;
+ case ARPHRD_ROSE:
+ return ROSE_ADDR_LEN;
+ case ARPHRD_NETROM:
+ return AX25_ADDR_LEN;
+ case ARPHRD_LOCALTLK:
+ return LTALK_ALEN;
+ default:
+ return 0;
+ }
+}
+
static long __tun_chr_ioctl(struct file *file, unsigned int cmd,
unsigned long arg, int ifreq_len)
{
@@ -3088,6 +3135,7 @@ static long __tun_chr_ioctl(struct file *file, unsigned int cmd,
break;
}
tun->dev->type = (int) arg;
+ tun->dev->addr_len = tun_get_addr_len(tun->dev->type);
netif_info(tun, drv, tun->dev, "linktype set to %d\n",
tun->dev->type);
call_netdevice_notifiers(NETDEV_POST_TYPE_CHANGE,
--
2.30.2


2021-04-06 09:44:11

by David Miller

[permalink] [raw]
Subject: Re: [PATCH] net: tun: set tun->dev->addr_len during TUNSETLINK processing

From: Phillip Potter <[email protected]>
Date: Mon, 5 Apr 2021 12:35:55 +0100

> When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
> to match the appropriate type, using new tun_get_addr_len utility function
> which returns appropriate address length for given type. Fixes a
> KMSAN-found uninit-value bug reported by syzbot at:
> https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51
>
> Reported-by: [email protected]
> Signed-off-by: Phillip Potter <[email protected]>
> ---
> drivers/net/tun.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 48 insertions(+)
>
> diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> index 978ac0981d16..56c26339ee3b 100644
> --- a/drivers/net/tun.c
> +++ b/drivers/net/tun.c
> @@ -69,6 +69,14 @@
> #include <linux/bpf.h>
> #include <linux/bpf_trace.h>
> #include <linux/mutex.h>
> +#include <linux/ieee802154.h>
> +#include <linux/if_ltalk.h>
> +#include <uapi/linux/if_fddi.h>
> +#include <uapi/linux/if_hippi.h>
> +#include <uapi/linux/if_fc.h>
> +#include <net/ax25.h>
> +#include <net/rose.h>
> +#include <net/6lowpan.h>
>
> #include <linux/uaccess.h>
> #include <linux/proc_fs.h>
> @@ -2925,6 +2933,45 @@ static int tun_set_ebpf(struct tun_struct *tun, struct tun_prog __rcu **prog_p,
> return __tun_set_ebpf(tun, prog_p, prog);
> }
>
> +/* Return correct value for tun->dev->addr_len based on tun->dev->type. */
> +static inline unsigned char tun_get_addr_len(unsigned short type)
> +{

Please do not use inline in foo.c files, let the compiler decide.

Thanks.

2021-04-06 19:40:15

by Phillip Potter

[permalink] [raw]
Subject: Re: [PATCH] net: tun: set tun->dev->addr_len during TUNSETLINK processing

On Mon, Apr 05, 2021 at 02:59:21PM -0700, David Miller wrote:
> From: Phillip Potter <[email protected]>
> Date: Mon, 5 Apr 2021 12:35:55 +0100
>
> > When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
> > to match the appropriate type, using new tun_get_addr_len utility function
> > which returns appropriate address length for given type. Fixes a
> > KMSAN-found uninit-value bug reported by syzbot at:
> > https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51
> >
> > Reported-by: [email protected]
> > Signed-off-by: Phillip Potter <[email protected]>
> > ---
> > drivers/net/tun.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++
> > 1 file changed, 48 insertions(+)
> >
> > diff --git a/drivers/net/tun.c b/drivers/net/tun.c
> > index 978ac0981d16..56c26339ee3b 100644
> > --- a/drivers/net/tun.c
> > +++ b/drivers/net/tun.c
> > @@ -69,6 +69,14 @@
> > #include <linux/bpf.h>
> > #include <linux/bpf_trace.h>
> > #include <linux/mutex.h>
> > +#include <linux/ieee802154.h>
> > +#include <linux/if_ltalk.h>
> > +#include <uapi/linux/if_fddi.h>
> > +#include <uapi/linux/if_hippi.h>
> > +#include <uapi/linux/if_fc.h>
> > +#include <net/ax25.h>
> > +#include <net/rose.h>
> > +#include <net/6lowpan.h>
> >
> > #include <linux/uaccess.h>
> > #include <linux/proc_fs.h>
> > @@ -2925,6 +2933,45 @@ static int tun_set_ebpf(struct tun_struct *tun, struct tun_prog __rcu **prog_p,
> > return __tun_set_ebpf(tun, prog_p, prog);
> > }
> >
> > +/* Return correct value for tun->dev->addr_len based on tun->dev->type. */
> > +static inline unsigned char tun_get_addr_len(unsigned short type)
> > +{
>
> Please do not use inline in foo.c files, let the compiler decide.
>
> Thanks.

Dear David,

Thank you for the feedback, I will resend.

Regards,
Phil

2021-04-07 10:27:14

by Eric Dumazet

[permalink] [raw]
Subject: Re: [PATCH] net: tun: set tun->dev->addr_len during TUNSETLINK processing



On 4/5/21 1:35 PM, Phillip Potter wrote:
> When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
> to match the appropriate type, using new tun_get_addr_len utility function
> which returns appropriate address length for given type. Fixes a
> KMSAN-found uninit-value bug reported by syzbot at:
> https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51
>
> Reported-by: [email protected]
> Signed-off-by: Phillip Potter <[email protected]>
> ---

Please give credits to people who helped.

You could have :

Suggested-by: Eric Dumazet <[email protected]>

Or
Diagnosed-by: Eric Dumazet <[email protected]>

Or at least CCed me :/

2021-04-07 10:29:44

by Phillip Potter

[permalink] [raw]
Subject: Re: [PATCH] net: tun: set tun->dev->addr_len during TUNSETLINK processing

On Tue, Apr 06, 2021 at 07:26:29PM +0200, Eric Dumazet wrote:
>
>
> On 4/5/21 1:35 PM, Phillip Potter wrote:
> > When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
> > to match the appropriate type, using new tun_get_addr_len utility function
> > which returns appropriate address length for given type. Fixes a
> > KMSAN-found uninit-value bug reported by syzbot at:
> > https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51
> >
> > Reported-by: [email protected]
> > Signed-off-by: Phillip Potter <[email protected]>
> > ---
>
> Please give credits to people who helped.
>
> You could have :
>
> Suggested-by: Eric Dumazet <[email protected]>
>
> Or
> Diagnosed-by: Eric Dumazet <[email protected]>
>
> Or at least CCed me :/
>

Dear Eric,

Please accept my apology for this oversight. It certainly wasn't
intentional on my part, and entirely down to inexperience. I will send
the patch again with Diagnosed-by.

Regards,
Phil