[v1] virtio-net: fix received length check in big packets

[PATCH net] virtio-net: fix received length check in big packets

Posted by Bui Quang Minh 3 months ago

Since commit 4959aebba8c0 ("virtio-net: use mtu size as buffer length
for big packets"), the allocated size for big packets is not
MAX_SKB_FRAGS * PAGE_SIZE anymore but depends on negotiated MTU. The
number of allocated frags for big packets is stored in
vi->big_packets_num_skbfrags. This commit fixes the received length
check corresponding to that change. The current incorrect check can lead
to NULL page pointer dereference in the below while loop when erroneous
length is received.

Fixes: 4959aebba8c0 ("virtio-net: use mtu size as buffer length for big packets")
Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
---
 drivers/net/virtio_net.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 5d674eb9a0f2..ead1cd2fb8af 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -823,7 +823,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 {
 	struct sk_buff *skb;
 	struct virtio_net_common_hdr *hdr;
-	unsigned int copy, hdr_len, hdr_padded_len;
+	unsigned int copy, hdr_len, hdr_padded_len, max_remaining_len;
 	struct page *page_to_free = NULL;
 	int tailroom, shinfo_size;
 	char *p, *hdr_p, *buf;
@@ -887,12 +887,16 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 	 * tries to receive more than is possible. This is usually
 	 * the case of a broken device.
 	 */
-	if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
+	BUG_ON(offset >= PAGE_SIZE);
+	max_remaining_len = (unsigned int)PAGE_SIZE - offset;
+	max_remaining_len += vi->big_packets_num_skbfrags * PAGE_SIZE;
+	if (unlikely(len > max_remaining_len)) {
 		net_dbg_ratelimited("%s: too much data\n", skb->dev->name);
 		dev_kfree_skb(skb);
+		give_pages(rq, page);
 		return NULL;
 	}
-	BUG_ON(offset >= PAGE_SIZE);
+
 	while (len) {
 		unsigned int frag_size = min((unsigned)PAGE_SIZE - offset, len);
 		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page, offset,
-- 
2.43.0

Re: [PATCH net] virtio-net: fix received length check in big packets

Posted by Jason Wang 3 months ago

On Sun, Jul 6, 2025 at 10:12 PM Bui Quang Minh <minhquangbui99@gmail.com> wrote:
>
> Since commit 4959aebba8c0 ("virtio-net: use mtu size as buffer length
> for big packets"), the allocated size for big packets is not
> MAX_SKB_FRAGS * PAGE_SIZE anymore but depends on negotiated MTU. The
> number of allocated frags for big packets is stored in
> vi->big_packets_num_skbfrags. This commit fixes the received length
> check corresponding to that change. The current incorrect check can lead
> to NULL page pointer dereference in the below while loop when erroneous
> length is received.
>
> Fixes: 4959aebba8c0 ("virtio-net: use mtu size as buffer length for big packets")
> Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
> ---
>  drivers/net/virtio_net.c | 10 +++++++---
>  1 file changed, 7 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 5d674eb9a0f2..ead1cd2fb8af 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -823,7 +823,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>  {
>         struct sk_buff *skb;
>         struct virtio_net_common_hdr *hdr;
> -       unsigned int copy, hdr_len, hdr_padded_len;
> +       unsigned int copy, hdr_len, hdr_padded_len, max_remaining_len;
>         struct page *page_to_free = NULL;
>         int tailroom, shinfo_size;
>         char *p, *hdr_p, *buf;
> @@ -887,12 +887,16 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>          * tries to receive more than is possible. This is usually
>          * the case of a broken device.
>          */
> -       if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
> +       BUG_ON(offset >= PAGE_SIZE);
> +       max_remaining_len = (unsigned int)PAGE_SIZE - offset;
> +       max_remaining_len += vi->big_packets_num_skbfrags * PAGE_SIZE;
> +       if (unlikely(len > max_remaining_len)) {
>                 net_dbg_ratelimited("%s: too much data\n", skb->dev->name);
>                 dev_kfree_skb(skb);
> +               give_pages(rq, page);

Should this be an independent fix?

>                 return NULL;
>         }
> -       BUG_ON(offset >= PAGE_SIZE);
> +
>         while (len) {
>                 unsigned int frag_size = min((unsigned)PAGE_SIZE - offset, len);
>                 skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page, offset,
> --
> 2.43.0
>

Thanks

Re: [PATCH net] virtio-net: fix received length check in big packets

Posted by Bui Quang Minh 3 months ago

On 7/7/25 10:48, Jason Wang wrote:
> On Sun, Jul 6, 2025 at 10:12 PM Bui Quang Minh <minhquangbui99@gmail.com> wrote:
>> Since commit 4959aebba8c0 ("virtio-net: use mtu size as buffer length
>> for big packets"), the allocated size for big packets is not
>> MAX_SKB_FRAGS * PAGE_SIZE anymore but depends on negotiated MTU. The
>> number of allocated frags for big packets is stored in
>> vi->big_packets_num_skbfrags. This commit fixes the received length
>> check corresponding to that change. The current incorrect check can lead
>> to NULL page pointer dereference in the below while loop when erroneous
>> length is received.
>>
>> Fixes: 4959aebba8c0 ("virtio-net: use mtu size as buffer length for big packets")
>> Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
>> ---
>>   drivers/net/virtio_net.c | 10 +++++++---
>>   1 file changed, 7 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index 5d674eb9a0f2..ead1cd2fb8af 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -823,7 +823,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>>   {
>>          struct sk_buff *skb;
>>          struct virtio_net_common_hdr *hdr;
>> -       unsigned int copy, hdr_len, hdr_padded_len;
>> +       unsigned int copy, hdr_len, hdr_padded_len, max_remaining_len;
>>          struct page *page_to_free = NULL;
>>          int tailroom, shinfo_size;
>>          char *p, *hdr_p, *buf;
>> @@ -887,12 +887,16 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>>           * tries to receive more than is possible. This is usually
>>           * the case of a broken device.
>>           */
>> -       if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
>> +       BUG_ON(offset >= PAGE_SIZE);
>> +       max_remaining_len = (unsigned int)PAGE_SIZE - offset;
>> +       max_remaining_len += vi->big_packets_num_skbfrags * PAGE_SIZE;
>> +       if (unlikely(len > max_remaining_len)) {
>>                  net_dbg_ratelimited("%s: too much data\n", skb->dev->name);
>>                  dev_kfree_skb(skb);
>> +               give_pages(rq, page);
> Should this be an independent fix?

I've realized this is not needed. In receive_big(), if page_to_skb() 
returns NULL then give_pages() is called. I'll remove this in next version.

Thanks,
Quang Minh.

>
>>                  return NULL;
>>          }
>> -       BUG_ON(offset >= PAGE_SIZE);
>> +
>>          while (len) {
>>                  unsigned int frag_size = min((unsigned)PAGE_SIZE - offset, len);
>>                  skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page, offset,
>> --
>> 2.43.0
>>
> Thanks
>

Re: [PATCH net] virtio-net: fix received length check in big packets

Posted by Bui Quang Minh 3 months ago

On 7/8/25 21:26, Bui Quang Minh wrote:
> On 7/7/25 10:48, Jason Wang wrote:
>> On Sun, Jul 6, 2025 at 10:12 PM Bui Quang Minh 
>> <minhquangbui99@gmail.com> wrote:
>>> Since commit 4959aebba8c0 ("virtio-net: use mtu size as buffer length
>>> for big packets"), the allocated size for big packets is not
>>> MAX_SKB_FRAGS * PAGE_SIZE anymore but depends on negotiated MTU. The
>>> number of allocated frags for big packets is stored in
>>> vi->big_packets_num_skbfrags. This commit fixes the received length
>>> check corresponding to that change. The current incorrect check can 
>>> lead
>>> to NULL page pointer dereference in the below while loop when erroneous
>>> length is received.
>>>
>>> Fixes: 4959aebba8c0 ("virtio-net: use mtu size as buffer length for 
>>> big packets")
>>> Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
>>> ---
>>>   drivers/net/virtio_net.c | 10 +++++++---
>>>   1 file changed, 7 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>>> index 5d674eb9a0f2..ead1cd2fb8af 100644
>>> --- a/drivers/net/virtio_net.c
>>> +++ b/drivers/net/virtio_net.c
>>> @@ -823,7 +823,7 @@ static struct sk_buff *page_to_skb(struct 
>>> virtnet_info *vi,
>>>   {
>>>          struct sk_buff *skb;
>>>          struct virtio_net_common_hdr *hdr;
>>> -       unsigned int copy, hdr_len, hdr_padded_len;
>>> +       unsigned int copy, hdr_len, hdr_padded_len, max_remaining_len;
>>>          struct page *page_to_free = NULL;
>>>          int tailroom, shinfo_size;
>>>          char *p, *hdr_p, *buf;
>>> @@ -887,12 +887,16 @@ static struct sk_buff *page_to_skb(struct 
>>> virtnet_info *vi,
>>>           * tries to receive more than is possible. This is usually
>>>           * the case of a broken device.
>>>           */
>>> -       if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
>>> +       BUG_ON(offset >= PAGE_SIZE);
>>> +       max_remaining_len = (unsigned int)PAGE_SIZE - offset;
>>> +       max_remaining_len += vi->big_packets_num_skbfrags * PAGE_SIZE;
>>> +       if (unlikely(len > max_remaining_len)) {
>>>                  net_dbg_ratelimited("%s: too much data\n", 
>>> skb->dev->name);
>>>                  dev_kfree_skb(skb);
>>> +               give_pages(rq, page);
>> Should this be an independent fix?
>
> I've realized this is not needed. In receive_big(), if page_to_skb() 
> returns NULL then give_pages() is called. I'll remove this in next 
> version.

It's a wrong fix actually. Calling give_pages twice will create bug.

>
>>
>>>                  return NULL;
>>>          }
>>> -       BUG_ON(offset >= PAGE_SIZE);
>>> +
>>>          while (len) {
>>>                  unsigned int frag_size = min((unsigned)PAGE_SIZE - 
>>> offset, len);
>>>                  skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, 
>>> page, offset,
>>> -- 
>>> 2.43.0
>>>
>> Thanks
>>
>

Re: [PATCH net] virtio-net: fix received length check in big packets

Posted by Jason Wang 3 months ago

On Mon, Jul 7, 2025 at 11:48 AM Jason Wang <jasowang@redhat.com> wrote:
>
> On Sun, Jul 6, 2025 at 10:12 PM Bui Quang Minh <minhquangbui99@gmail.com> wrote:
> >
> > Since commit 4959aebba8c0 ("virtio-net: use mtu size as buffer length
> > for big packets"), the allocated size for big packets is not
> > MAX_SKB_FRAGS * PAGE_SIZE anymore but depends on negotiated MTU. The
> > number of allocated frags for big packets is stored in
> > vi->big_packets_num_skbfrags. This commit fixes the received length
> > check corresponding to that change. The current incorrect check can lead
> > to NULL page pointer dereference in the below while loop when erroneous
> > length is received.
> >
> > Fixes: 4959aebba8c0 ("virtio-net: use mtu size as buffer length for big packets")
> > Signed-off-by: Bui Quang Minh <minhquangbui99@gmail.com>
> > ---
> >  drivers/net/virtio_net.c | 10 +++++++---
> >  1 file changed, 7 insertions(+), 3 deletions(-)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 5d674eb9a0f2..ead1cd2fb8af 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -823,7 +823,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
> >  {
> >         struct sk_buff *skb;
> >         struct virtio_net_common_hdr *hdr;
> > -       unsigned int copy, hdr_len, hdr_padded_len;
> > +       unsigned int copy, hdr_len, hdr_padded_len, max_remaining_len;
> >         struct page *page_to_free = NULL;
> >         int tailroom, shinfo_size;
> >         char *p, *hdr_p, *buf;
> > @@ -887,12 +887,16 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
> >          * tries to receive more than is possible. This is usually
> >          * the case of a broken device.
> >          */
> > -       if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
> > +       BUG_ON(offset >= PAGE_SIZE);
> > +       max_remaining_len = (unsigned int)PAGE_SIZE - offset;
> > +       max_remaining_len += vi->big_packets_num_skbfrags * PAGE_SIZE;
> > +       if (unlikely(len > max_remaining_len)) {
> >                 net_dbg_ratelimited("%s: too much data\n", skb->dev->name);
> >                 dev_kfree_skb(skb);
> > +               give_pages(rq, page);
>
> Should this be an independent fix?

Btw, the page_to_skb() is really kind of messed up right now that it
is used by both big and mergeable. We may need to consider splitting
the logic in the future.

Thanks

>
> >                 return NULL;
> >         }
> > -       BUG_ON(offset >= PAGE_SIZE);
> > +
> >         while (len) {
> >                 unsigned int frag_size = min((unsigned)PAGE_SIZE - offset, len);
> >                 skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, page, offset,
> > --
> > 2.43.0
> >
>
> Thanks