[PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[Wang Yufen]

Resurrect ubsan overflow checks and ubsan report this warning,
fix it by change the variable [length] type to size_t.

UBSAN: signed-integer-overflow in net/ipv6/ip6_output.c:1489:19
2147479552 + 8567 cannot be represented in type 'int'
CPU: 0 PID: 253 Comm: err Not tainted 5.16.0+ #1
Hardware name: linux,dummy-virt (DT)
Call trace:
  dump_backtrace+0x214/0x230
  show_stack+0x30/0x78
  dump_stack_lvl+0xf8/0x118
  dump_stack+0x18/0x30
  ubsan_epilogue+0x18/0x60
  handle_overflow+0xd0/0xf0
  __ubsan_handle_add_overflow+0x34/0x44
  __ip6_append_data.isra.48+0x1598/0x1688
  ip6_append_data+0x128/0x260
  udpv6_sendmsg+0x680/0xdd0
  inet6_sendmsg+0x54/0x90
  sock_sendmsg+0x70/0x88
  ____sys_sendmsg+0xe8/0x368
  ___sys_sendmsg+0x98/0xe0
  __sys_sendmmsg+0xf4/0x3b8
  __arm64_sys_sendmmsg+0x34/0x48
  invoke_syscall+0x64/0x160
  el0_svc_common.constprop.4+0x124/0x300
  do_el0_svc+0x44/0xc8
  el0_svc+0x3c/0x1e8
  el0t_64_sync_handler+0x88/0xb0
  el0t_64_sync+0x16c/0x170

Changes since v1:
-Change the variable [length] type to unsigned, as Eric Dumazet suggested.
Changes since v2:
-Don't change exthdrlen type in ip6_make_skb, as Paolo Abeni suggested.
Changes since v3:
-Don't change ulen type in udpv6_sendmsg and l2tp_ip6_sendmsg, as
Jakub Kicinski suggested.

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Yufen <wangyufen@huawei.com>
---
 include/net/ipv6.h    | 4 ++--
 net/ipv6/ip6_output.c | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/include/net/ipv6.h b/include/net/ipv6.h
index 5b38bf1a586b..de9dcc5652c4 100644
--- a/include/net/ipv6.h
+++ b/include/net/ipv6.h
@@ -1063,7 +1063,7 @@ int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
 int ip6_append_data(struct sock *sk,
 		    int getfrag(void *from, char *to, int offset, int len,
 				int odd, struct sk_buff *skb),
-		    void *from, int length, int transhdrlen,
+		    void *from, size_t length, int transhdrlen,
 		    struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
 		    struct rt6_info *rt, unsigned int flags);
 
@@ -1079,7 +1079,7 @@ struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
 struct sk_buff *ip6_make_skb(struct sock *sk,
 			     int getfrag(void *from, char *to, int offset,
 					 int len, int odd, struct sk_buff *skb),
-			     void *from, int length, int transhdrlen,
+			     void *from, size_t length, int transhdrlen,
 			     struct ipcm6_cookie *ipc6,
 			     struct rt6_info *rt, unsigned int flags,
 			     struct inet_cork_full *cork);
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 4081b12a01ff..77e3f5970ce4 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -1450,7 +1450,7 @@ static int __ip6_append_data(struct sock *sk,
 			     struct page_frag *pfrag,
 			     int getfrag(void *from, char *to, int offset,
 					 int len, int odd, struct sk_buff *skb),
-			     void *from, int length, int transhdrlen,
+			     void *from, size_t length, int transhdrlen,
 			     unsigned int flags, struct ipcm6_cookie *ipc6)
 {
 	struct sk_buff *skb, *skb_prev = NULL;
@@ -1798,7 +1798,7 @@ static int __ip6_append_data(struct sock *sk,
 int ip6_append_data(struct sock *sk,
 		    int getfrag(void *from, char *to, int offset, int len,
 				int odd, struct sk_buff *skb),
-		    void *from, int length, int transhdrlen,
+		    void *from, size_t length, int transhdrlen,
 		    struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
 		    struct rt6_info *rt, unsigned int flags)
 {
@@ -1995,7 +1995,7 @@ EXPORT_SYMBOL_GPL(ip6_flush_pending_frames);
 struct sk_buff *ip6_make_skb(struct sock *sk,
 			     int getfrag(void *from, char *to, int offset,
 					 int len, int odd, struct sk_buff *skb),
-			     void *from, int length, int transhdrlen,
+			     void *from, size_t length, int transhdrlen,
 			     struct ipcm6_cookie *ipc6, struct rt6_info *rt,
 			     unsigned int flags, struct inet_cork_full *cork)
 {
-- 
2.25.1

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[Paolo Abeni]

On Wed, 2022-06-01 at 16:48 +0800, Wang Yufen wrote:
> Resurrect ubsan overflow checks and ubsan report this warning,
> fix it by change the variable [length] type to size_t.
> 
> UBSAN: signed-integer-overflow in net/ipv6/ip6_output.c:1489:19
> 2147479552 + 8567 cannot be represented in type 'int'
> CPU: 0 PID: 253 Comm: err Not tainted 5.16.0+ #1
> Hardware name: linux,dummy-virt (DT)
> Call trace:
>   dump_backtrace+0x214/0x230
>   show_stack+0x30/0x78
>   dump_stack_lvl+0xf8/0x118
>   dump_stack+0x18/0x30
>   ubsan_epilogue+0x18/0x60
>   handle_overflow+0xd0/0xf0
>   __ubsan_handle_add_overflow+0x34/0x44
>   __ip6_append_data.isra.48+0x1598/0x1688
>   ip6_append_data+0x128/0x260
>   udpv6_sendmsg+0x680/0xdd0
>   inet6_sendmsg+0x54/0x90
>   sock_sendmsg+0x70/0x88
>   ____sys_sendmsg+0xe8/0x368
>   ___sys_sendmsg+0x98/0xe0
>   __sys_sendmmsg+0xf4/0x3b8
>   __arm64_sys_sendmmsg+0x34/0x48
>   invoke_syscall+0x64/0x160
>   el0_svc_common.constprop.4+0x124/0x300
>   do_el0_svc+0x44/0xc8
>   el0_svc+0x3c/0x1e8
>   el0t_64_sync_handler+0x88/0xb0
>   el0t_64_sync+0x16c/0x170
> 
> Changes since v1:
> -Change the variable [length] type to unsigned, as Eric Dumazet suggested.
> Changes since v2:
> -Don't change exthdrlen type in ip6_make_skb, as Paolo Abeni suggested.
> Changes since v3:
> -Don't change ulen type in udpv6_sendmsg and l2tp_ip6_sendmsg, as
> Jakub Kicinski suggested.

I'm sorry for the multiple incremental feedback on this patch. It's
somewhat tricky.

AFAICS Jakub mentioned only udpv6_sendmsg(). In l2tp_ip6_sendmsg() we
can have an overflow:

        int transhdrlen = 4; /* zero session-id */
        int ulen = len + transhdrlen;

when len >= INT_MAX - 4. That will be harmless, but I guess it could
still trigger a noisy UBSAN splat. 

Paolo

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[Jakub Kicinski]

On Thu, 02 Jun 2022 12:38:10 +0200 Paolo Abeni wrote:
> I'm sorry for the multiple incremental feedback on this patch. It's
> somewhat tricky.
> 
> AFAICS Jakub mentioned only udpv6_sendmsg(). In l2tp_ip6_sendmsg() we
> can have an overflow:
> 
>         int transhdrlen = 4; /* zero session-id */
>         int ulen = len + transhdrlen;
> 
> when len >= INT_MAX - 4. That will be harmless, but I guess it could
> still trigger a noisy UBSAN splat. 

Good point, I wonder if that's a separate issue. Should we
follow what UDP does and subtract the transhdr from the max?
My gut feeling is that stricter checks are cleaner than just 
bumping variable sizes.

diff --git a/net/l2tp/l2tp_ip6.c b/net/l2tp/l2tp_ip6.c
index c6ff8bf9b55f..9dbd801ddb98 100644
--- a/net/l2tp/l2tp_ip6.c
+++ b/net/l2tp/l2tp_ip6.c
@@ -504,14 +504,15 @@ static int l2tp_ip6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
        struct ipcm6_cookie ipc6;
        int addr_len = msg->msg_namelen;
        int transhdrlen = 4; /* zero session-id */
-       int ulen = len + transhdrlen;
+       int ulen;
        int err;
 
        /* Rough check on arithmetic overflow,
         * better check is made in ip6_append_data().
         */
-       if (len > INT_MAX)
+       if (len > INT_MAX - transhdrlen)
                return -EMSGSIZE;
+       ulen = len + transhdrlen;
 
        /* Mirror BSD error message compatibility */
        if (msg->msg_flags & MSG_OOB)

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[Paolo Abeni]

On Thu, 2022-06-02 at 09:02 -0700, Jakub Kicinski wrote:
> On Thu, 02 Jun 2022 12:38:10 +0200 Paolo Abeni wrote:
> > I'm sorry for the multiple incremental feedback on this patch. It's
> > somewhat tricky.
> > 
> > AFAICS Jakub mentioned only udpv6_sendmsg(). In l2tp_ip6_sendmsg() we
> > can have an overflow:
> > 
> >         int transhdrlen = 4; /* zero session-id */
> >         int ulen = len + transhdrlen;
> > 
> > when len >= INT_MAX - 4. That will be harmless, but I guess it could
> > still trigger a noisy UBSAN splat. 
> 
> Good point, I wonder if that's a separate issue. Should we
> follow what UDP does and subtract the transhdr from the max?
> My gut feeling is that stricter checks are cleaner than just 
> bumping variable sizes.
> 
> diff --git a/net/l2tp/l2tp_ip6.c b/net/l2tp/l2tp_ip6.c
> index c6ff8bf9b55f..9dbd801ddb98 100644
> --- a/net/l2tp/l2tp_ip6.c
> +++ b/net/l2tp/l2tp_ip6.c
> @@ -504,14 +504,15 @@ static int l2tp_ip6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
>         struct ipcm6_cookie ipc6;
>         int addr_len = msg->msg_namelen;
>         int transhdrlen = 4; /* zero session-id */
> -       int ulen = len + transhdrlen;
> +       int ulen;
>         int err;
>  
>         /* Rough check on arithmetic overflow,
>          * better check is made in ip6_append_data().
>          */
> -       if (len > INT_MAX)
> +       if (len > INT_MAX - transhdrlen)
>                 return -EMSGSIZE;
> +       ulen = len + transhdrlen;
>  
>         /* Mirror BSD error message compatibility */
>         if (msg->msg_flags & MSG_OOB)
> 
LGTM. Imho this can even land in a separated patch (whatever is easier)

Thanks!

Paolo

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[wangyufen]

在 2022/6/3 16:58, Paolo Abeni 写道:
> On Thu, 2022-06-02 at 09:02 -0700, Jakub Kicinski wrote:
>> On Thu, 02 Jun 2022 12:38:10 +0200 Paolo Abeni wrote:
>>> I'm sorry for the multiple incremental feedback on this patch. It's
>>> somewhat tricky.
>>>
>>> AFAICS Jakub mentioned only udpv6_sendmsg(). In l2tp_ip6_sendmsg() we
>>> can have an overflow:
>>>
>>>          int transhdrlen = 4; /* zero session-id */
>>>          int ulen = len + transhdrlen;
>>>
>>> when len >= INT_MAX - 4. That will be harmless, but I guess it could
>>> still trigger a noisy UBSAN splat.
>> Good point, I wonder if that's a separate issue. Should we
>> follow what UDP does and subtract the transhdr from the max?
>> My gut feeling is that stricter checks are cleaner than just
>> bumping variable sizes.
>>
>> diff --git a/net/l2tp/l2tp_ip6.c b/net/l2tp/l2tp_ip6.c
>> index c6ff8bf9b55f..9dbd801ddb98 100644
>> --- a/net/l2tp/l2tp_ip6.c
>> +++ b/net/l2tp/l2tp_ip6.c
>> @@ -504,14 +504,15 @@ static int l2tp_ip6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
>>          struct ipcm6_cookie ipc6;
>>          int addr_len = msg->msg_namelen;
>>          int transhdrlen = 4; /* zero session-id */
>> -       int ulen = len + transhdrlen;
>> +       int ulen;
>>          int err;
>>   
>>          /* Rough check on arithmetic overflow,
>>           * better check is made in ip6_append_data().
>>           */
>> -       if (len > INT_MAX)
>> +       if (len > INT_MAX - transhdrlen)
>>                  return -EMSGSIZE;
>> +       ulen = len + transhdrlen;
>>   
>>          /* Mirror BSD error message compatibility */
>>          if (msg->msg_flags & MSG_OOB)
>>
> LGTM. Imho this can even land in a separated patch (whatever is easier)

Thanks for all the feedback.
So, Jakub will send a new patch to fix the l2tp_ip6_sendmsg issue?

Thanks.

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[Jakub Kicinski]

On Mon, 6 Jun 2022 10:03:27 +0800 wangyufen wrote:
> > LGTM. Imho this can even land in a separated patch (whatever is easier)  
> 
> Thanks for all the feedback.
> So, Jakub will send a new patch to fix the l2tp_ip6_sendmsg issue?

That was just a suggestion, if possible I'd prefer if you double
checked the analysis, tested that or similar code, wrote a commit
message etc. and sent it.

Re: [PATCH net-next v4] ipv6: Fix signed integer overflow in __ip6_append_data

[wangyufen]

在 2022/6/6 22:24, Jakub Kicinski 写道:
> On Mon, 6 Jun 2022 10:03:27 +0800 wangyufen wrote:
>>> LGTM. Imho this can even land in a separated patch (whatever is easier)
>> Thanks for all the feedback.
>> So, Jakub will send a new patch to fix the l2tp_ip6_sendmsg issue?
> That was just a suggestion, if possible I'd prefer if you double
> checked the analysis, tested that or similar code, wrote a commit
> message etc. and sent it.
> .
OK, thanks. will send in v5.