[PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[Florian Fainelli]

On 64-bits hosts, napi_gro_cb is 48 bytes, which is exactly the size of
skb->cb[], while on 32-bits hosts it is 36 bytes, but if it were to
grow, we would not be catching a size inflation as we should.

Make sure that we have enough room for a napi_gro_cb to be hosted in
skb->cb[], and put this build-time assertion in skb_gro_reset_offset()
since this function is invoked by the GRO layers entry points.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
 net/core/dev.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/net/core/dev.c b/net/core/dev.c
index 4870c3556a5a..20bc82604b75 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4115,6 +4115,8 @@ static void skb_gro_reset_offset(struct sk_buff *skb)
 	const struct skb_shared_info *pinfo = skb_shinfo(skb);
 	const skb_frag_t *frag0 = &pinfo->frags[0];
 
+	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct napi_gro_cb));
+
 	NAPI_GRO_CB(skb)->data_offset = 0;
 	NAPI_GRO_CB(skb)->frag0 = NULL;
 	NAPI_GRO_CB(skb)->frag0_len = 0;
-- 
2.1.0

Re: [PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[Tom Herbert]

On Fri, Jul 31, 2015 at 11:51 AM, Florian Fainelli <f.fainelli@gmail.com> wrote:
> On 64-bits hosts, napi_gro_cb is 48 bytes, which is exactly the size of
> skb->cb[], while on 32-bits hosts it is 36 bytes, but if it were to
> grow, we would not be catching a size inflation as we should.
>
> Make sure that we have enough room for a napi_gro_cb to be hosted in
> skb->cb[], and put this build-time assertion in skb_gro_reset_offset()
> since this function is invoked by the GRO layers entry points.
>
> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> ---
>  net/core/dev.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/net/core/dev.c b/net/core/dev.c
> index 4870c3556a5a..20bc82604b75 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -4115,6 +4115,8 @@ static void skb_gro_reset_offset(struct sk_buff *skb)
>         const struct skb_shared_info *pinfo = skb_shinfo(skb);
>         const skb_frag_t *frag0 = &pinfo->frags[0];
>
> +       BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct napi_gro_cb));
> +
Maybe sizeof(struct napi_gro_cb) > sizeof(skb->cb) since this about
napi_gro_cb being too big, not cb being too small.

>         NAPI_GRO_CB(skb)->data_offset = 0;
>         NAPI_GRO_CB(skb)->frag0 = NULL;
>         NAPI_GRO_CB(skb)->frag0_len = 0;
> --
> 2.1.0
>
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[Florian Westphal]

Florian Fainelli <f.fainelli@gmail.com> wrote:
> On 64-bits hosts, napi_gro_cb is 48 bytes, which is exactly the size of
> skb->cb[], while on 32-bits hosts it is 36 bytes, but if it were to
> grow, we would not be catching a size inflation as we should.
> ---
>  net/core/dev.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/net/core/dev.c b/net/core/dev.c
> index 4870c3556a5a..20bc82604b75 100644
> --- a/net/core/dev.c
> +++ b/net/core/dev.c
> @@ -4115,6 +4115,8 @@ static void skb_gro_reset_offset(struct sk_buff *skb)
>  	const struct skb_shared_info *pinfo = skb_shinfo(skb);
>  	const skb_frag_t *frag0 = &pinfo->frags[0];
>  
> +	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct napi_gro_cb));

commit c3c7c254b2e8cd99b0adf288c2a1bddacd7ba255
already added a check for this.

Re: [PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[David Miller]

From: Florian Westphal <fw@strlen.de>
Date: Fri, 31 Jul 2015 22:24:03 +0200

> Florian Fainelli <f.fainelli@gmail.com> wrote:
>> On 64-bits hosts, napi_gro_cb is 48 bytes, which is exactly the size of
>> skb->cb[], while on 32-bits hosts it is 36 bytes, but if it were to
>> grow, we would not be catching a size inflation as we should.
>> ---
>>  net/core/dev.c | 2 ++
>>  1 file changed, 2 insertions(+)
>> 
>> diff --git a/net/core/dev.c b/net/core/dev.c
>> index 4870c3556a5a..20bc82604b75 100644
>> --- a/net/core/dev.c
>> +++ b/net/core/dev.c
>> @@ -4115,6 +4115,8 @@ static void skb_gro_reset_offset(struct sk_buff *skb)
>>  	const struct skb_shared_info *pinfo = skb_shinfo(skb);
>>  	const skb_frag_t *frag0 = &pinfo->frags[0];
>>  
>> +	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct napi_gro_cb));
> 
> commit c3c7c254b2e8cd99b0adf288c2a1bddacd7ba255
> already added a check for this.

Indeed.

Re: [PATCH RFC net-next] net: Assert napi_gro_cb size against skb->cb[]

[Florian Fainelli]

On 31/07/15 13:44, David Miller wrote:
> From: Florian Westphal <fw@strlen.de>
> Date: Fri, 31 Jul 2015 22:24:03 +0200
> 
>> Florian Fainelli <f.fainelli@gmail.com> wrote:
>>> On 64-bits hosts, napi_gro_cb is 48 bytes, which is exactly the size of
>>> skb->cb[], while on 32-bits hosts it is 36 bytes, but if it were to
>>> grow, we would not be catching a size inflation as we should.
>>> ---
>>>  net/core/dev.c | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/net/core/dev.c b/net/core/dev.c
>>> index 4870c3556a5a..20bc82604b75 100644
>>> --- a/net/core/dev.c
>>> +++ b/net/core/dev.c
>>> @@ -4115,6 +4115,8 @@ static void skb_gro_reset_offset(struct sk_buff *skb)
>>>  	const struct skb_shared_info *pinfo = skb_shinfo(skb);
>>>  	const skb_frag_t *frag0 = &pinfo->frags[0];
>>>  
>>> +	BUILD_BUG_ON(sizeof(skb->cb) < sizeof(struct napi_gro_cb));
>>
>> commit c3c7c254b2e8cd99b0adf288c2a1bddacd7ba255
>> already added a check for this.
> 
> Indeed.

Whoops, missed that part, thanks Florian!
-- 
Florian